We had a super busy week so The Format is rolling out slightly late…

It’ll be worth it though because this week has been a big one.

Other than the HUGE releases by OpenAI… we’ll be looking at:

• How token prices are killing AI products 🧛‍♀️

• Grok’s unfair advantage… an AI bot that strips for you? 🤔

• The two sides of Zuck ❓

• How you can fly/walk/drive around your AI world 😍

Anyway let’s jump in!

Introducing GPT-5 and gpt-oss 👀

We had a double drop from OpenAI this week. GPT-5 and gpt-oss (yes one is upper case and the other lower, I don’t know why).

There has been a lot of public response to both.

Let’s start with GPT-5:

It’s definitely an upgrade, and the way they have unified all the models into the ‘GPT-5’ model is great. But after waiting for over 2 years, people were expecting and ‘Apple moment’.

And well… the general consensus is that we didn’t get one.

My two biggest takeaways from the long release video were:

• Everyone who is on the free tier will now get GPT-5, for the people who don’t really know/care what model they are using, it’s going to feel a whole lot smarter

• They briefly mentioned an integration with Gmail and Google Calendar. This feels like a huge thing, and with some more tactical integrations the personal context the models will have, and actions they will be able to take will make them exponentially more useful.

Now onto gpt-oss:

OpenAI also released 2 impressive open source models. The first time they’ve been open in quite a long time.

These models aren’t quite state of the art, but they aren’t too far off, and given the fact that most of the other top open models were built by Chinese companies, this release is a big win for American open source.

The fact that a big name like OpenAI has released these models will give so many enterprise customers the confidence to move towards a privacy focused approach.

Our experiments internally have already begun…

Tokens are getting more expensive

A lot of AI companies and AI-based products have fatal flaws with their pricing structure. It’s a race to the bottom, and it’s causing a blood bath.

When they came up with their pricing model it was under the following assumptions…

• year 1: break even at $20/month

• year 2: 90% margins as compute drops 10x

• year 3: yacht shopping

There’s a problem with this though.

Although the margins have dropped exactly as expected, customers are expecting the latest state of the art models.

And these models are actually costing significantly more due to ‘thinking’ time and the vast amount of additional tokens that are required to achieve this.

Companies aren’t making huge margins, they’re actually making huge losses on power users, it’s just not sustainable…

The piece does explore quite a few interesting solutions to this problem though.

My favourite is Replit’s approach: complete vertical integration.

By making sure that they have infrastructure covering every layer of the stack, and importantly that they make a good margin at each layer, they are able to essentially use the AI development part of their platform as a ‘marketing budget’.

Plus, as a user the switching costs become much, much higher, since you have to switch the whole stack.

It’s genius… and it’s working.

The World Becomes Playable

When I say ‘Think about AI” you almost certainly start picturing a large language model like ChatGPT, then maybe a voice or image model.

I bet your brain didn’t imagine a world model.

Wait… what is a world model?

Well this week Google’s Deepmind team released a paper on demo’s for Genie 3. A model that can simulate a 720p world for you based off a prompt/image/video, then in real-time you can navigate around it and interact with it. Mad. 🤯

The most impressive things about the model in my opinion:

• The physics… this model acts as if it has a built in real-time 3D physics engine (something Unreal Engine has spent billions on). Everything from the dynamic reflection in water to complex motion, it’s wild.

• The memory… although it’s currently only minutes long, it’s crazy being able to interact with an object e.g. paint on a wall or open a door, leave, go on a short adventure, come back and the world remains exactly as you left it

• The range… whether it’s exploring the world from the first person perspective of a bird, the third down perspective of a Mars Rover or the top down perspective of an 8-bit strategy game. The model can create almost any world, with any aesthetic and from any angle.

Will GTA 6 be AI generated?

Probably not… but GTA 7 almost certainly will be.

Other articles + videos we’ve found interesting this week…

Manus is launching ‘Wide Research’ that spins up 100+ agents to scour the web for you → The concept of running parallel agents at once and then bringing together some/all the results is something that we are only going to see more of over the next months and years. We’ve been experimenting internally… 👀

In Search of Native Business Models → In many ‘AI products’ it seems like the line between “tool” and “worker” is disappearing. For now usage-based pricing is king, but it’s only time before a new (hybrid?) pricing model will rise.

How a ‘vibe working’ approach at Genspark tripled ARR growth → In what they call ‘vibe working’, Genspark have implemented an AI-native workforce from the ground up. This is allowing them to move, pivot and grow incredibly fast.

When will ‘AI-native’ become the only viable way for teams to work?

Zuck’s note on Personal Superintelligence → Rather than industries replacing human work directly, Zuck discusses how he sees AI as a deeply personal assistant that knows your goals, helps you create, connect, and grow, and empowers billions to shape the future on their own terms.

This seems like a lofty contrast to their current business model with Meta pumping on the news that ‘video screen time’ is up 20% thanks to their curated doom scrolling feeds. We shall see.

But I don’t expect self-actualisation to come via Instagram Reels.

xAI Launches Companions That Can Engage in NSFW Chats → Premium X users can now chat to anime-styled ‘companions’. They have strong and unfiltered personalities, are very fun to talk too (if you’re into that), and you can be very intimate with them…

Ethically questionable? YES. But the strategic data-acquisition play makes a lot of sense. People will talk to it in a way they wouldn’t with Grok.

Means and Meaning → Whilst technology may be rapidly giving us more means, it doesn’t help with the problem of meaning. This essay by Packy McCormick dives into what happens as humans are abstracted away from productive work and how this is/will impact attention, meaning and life.

OpenAI validates the open ecosystem (finally) → OpenAI seem to be playing strategic games yet again. They have finally released their long anticipated open source model. But why? How does it benefit them?

One take is that they have released this model to undercut their competitors (Google, Anthropic etc.) and then they are aiming to frame GPT-5 as the state of the art go to model for everything else.

The battle for AI market share continues… 👀

What we’ve been up to…

1. We’ve had a little design overhaul. 🎨

Out with the black and yellow dark mode. In with purple and light mode.

So far feedback has been very positive!!

I see we have new UI refresh 😄 just dropped in to say I love it

The new dashboard looks amazing in open format! 🔥

This is just the start though, there’s a much bigger design and brand refresh in the works… keep your eyes peeled. 👀

2. Andy spent some time walking through the dashboard 💪

Below he shows some of the questions you can ask your Community Assistant, walks through a sentiment report and sends some rewards 🎁

3. We 💛 Pods

Internally we’ve slightly evolved how we work, and we’re now using a system where we have multiple pods.

Without diving into it too deeply what this should mean is that you’re going to see very rapid iterations of the product and assistant (at least 1 per week).

We’re super excited!!

4. We’ve written a CommunityOS Playbook. 🧠

If you’re interested and want access just let us know!

And that’s all for this week.

Have a great weekend, Dan and the Open Format team 👋🏽

A couple of asks…

• If you know a community that would be interested in getting a sentiment report for their Discord or Telegram just drop us a message. We’ll get it to you in no time. 🤘

• Check out our website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

Today we’ll be looking at:

• The curious case of the AI lawyer 🕵️‍♂️

• Death to the product. Long live distribution? 👀

• Finding your place in the workflow 🤩

• Everything you need to know about AI agents 💪

Anyway let’s jump in…

Surprise: you’re a custom workflow consultancy now.

Every time I scroll through Linkedin I see about 5 posts telling me that I NEED to clone this exact 55 step AI automation because it’ll save me 30 hours per day.

I get it. Plug and play AI automation’s can have value to some people, but these posts also make me laugh. The automations are one-size fits all, and that one-size never fits the tools and workflows we have at Open Format, why would they?

This is exactly the problem, or rather the opportunity, that this post talks about. The real value comes from creating something that can live in the messiness, in the complexity, in the many shapes and sizes of companies that exist today.

When you find a way to get through that messy complexity you can create a really deep and powerful integration.

You can create workflows and products that are impossible to rip out.

Systems that sit right at the heart (and through all the limbs) of the company.

There is a whole trend in AI and SaaS at the moment of ‘vertical integration’ and this is why. Because it comes with a whole tonne of potential value.

CommunityOS → Vertical integration for communities? 👀

Ship the wrapper. Ride the wave.

After a highly triggering first line: “Europe’s newest unicorn can be rebuilt in 75 minutes”, this post actually goes on to make some really interesting points about the state of AI products at the moment and the way to ‘win’.

The key takeaway, as models improve (and fast) there is way less space for technological differentiation between products. There are very few clear market leaders based on performance alone.

This means that the most important metrics of success become distribution and mindshare.

You become the de facto product that people use, not because you’re by far the best, but because you’re the product everyone associates with a vertical or a task.

ChatGPT have done this with commercial LLM’s

Lovable have done this with vibe coding.

Icon has done this with AI ads.

This isn’t saying that products no longer need to be good, they absolutely do.

It’s just that being the best purely on product performance no longer seems to be a viable product moat.

Principles of Building AI Agents

AI agents are everywhere right now, and they’re only going to continue proliferating until there’s a trillion of them. (only half joking…)

The whole topic of AI agents feels rather daunting, they can somehow remember things, do things, talk to each other? It’s a lot.

But once you look behind the curtains the agent is actually made up of a few quite simple parts.

Given the importance that these agents will play over the next 1,5 and 10 years, it feels like these primitives are something that everyone should understand.

And this short book by Mastra, an AI agent framework (so they certainly know their stuff) is a great and practical place to get started with learning and building!

For context, I read it in just over an hour, and I’m a slow reader.

Other articles + videos we’ve found interesting this week…

Knowledge Graphs in Litigation Agents → This video discusses the architecture of an AI agent specifically designed to assist lawyers. The most interesting part was their AI UX: they balanced 1:1 conversation (the AI UX we’re all familiar with) alongside AI generated detailed reports and summarisation.

Huge value was created from this hybrid approach… we’re taking a very similar approach but for communities. 👀

Swedish unicorn Lovable crosses the $100M ARR milestone → 2025 has been the year of the vibe coder. Lovable. Bolt. v0. Base44. There are countless vibe coding platforms, and there’s a very clear market, with Lovable alone having an impressive 180k paying subscribers. After Lovable’s recent $1.8b valuation, there will only be more products rushing to market. Saturation coming soon?

Reinforcement Learning the right way → As usual we have a hot take from Andrej Karparthy. This time discussing RL and how despite the fact that current RL can achieve a huge amount, to really take it to the next level we will need a ‘reflection stage’. This would mean that over time the lessons learned could be added to the AI’s system prompt and help it improve more globally.

Everything is Context Engineering → This 1 hour mastraclass covers just about everything you need to know on AI agent memory, delivered by the awesome Mastra team.

What we’ve been up to…

Over the past few weeks everyone in the Open Format team has had one thing on their mind…

Next week we’re shipping a tonne of new features and we’re finally going to start charging for the Pro tier of our product.

I don’t want to give too much away, but some of the features will include:

🧠 A community brain that learns as you chat to it: your goals, audience, preferences and just about everything else

🐙 A community assistant that you can chat to to find out more about your community, to co-create campaigns, to write content and so much more…

🕵️‍♂️ Reports on your moderators performance, their speed to reply, the quality of their responses etc.

Here’s a video I made to bring it to life…

We’ve also been generating sentiment reports for communities who want a taste of what our CommunityOS has to offer before they get started.

Here’s a video (made by our newest team member, Leo) that shows off the amazing sentiment in the Turbo community!

If you know a community that would be interested in getting a sentiment report for their Discord or Telegram just drop us a message. We’ll get it to you in no time. 🤘

And that’s all for this week.

Have a great weekend, Dan and the Open Format team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

  

Why?

For EthCC: a week of connecting with many, many, many interesting people from the web3 and AI ecosystems, deepening relationships with our partners (huge shout out to Aurora, Near and Coineasy) and aligning as a team.

We learned a lot, had a tonne of very useful conversations and attended some very interesting panels.

Here are some of our key reflections from the trip:

🚀 The product is resonating - and fast

One of the most encouraging signs was how strongly our product and positioning landed. We had an incredible ‘conversation rate’. Literally about 60–80% of the people we spoke to wanted the product for their project or community.

Thanks to a script Andy cooked up we were able to generate super insightful community sentiment reports for any community in just a couple of minutes. Seeing the reactions to these was one of the best parts of the trip, some minds were definitely blown and a lot of people were very interested and wanted more.

It was also very interesting to see how often people would mention ‘sentiment’, even before we introduced what we were building - a signal that the language we’re using is starting to resonate. Deeply.

💡 Strong alignment and team spirit

EthCC was as much about strengthening our team as it was about showcasing our product. As Dan put it, “the team meeting we had in Cannes was probably the best we’ve ever had - we left more aligned than ever and we’re very excited about what’s ahead.”

For newer team members, like Leo, it was also a great way to ask questions, get to know everyone properly and feel the energy of what we’re building.

🤝 Connections & partnerships

We had some fantastic conversations with friends and partners like Aurora, Beacon Protocol and PSG, as well as with countless new faces from across the ecosystem.

The optimism of the space was palpable, and the push toward privacy, AI integration, and meaningful community engagement aligned strongly with what we’re building.

🧭 Areas to improve

While EthCC was a success, it also highlighted areas where we can (and must) improve. For example, we noticed that phrases like “rewards” didn’t always land well with certain audiences, particularly the DeFi crowd, where “capital allocation” resonated better.

Once people got what we’re building, they really got it (and usually wanted it), but it sometimes took a while to get to that point. We’re rapidly improving how we communicate what we’re building though, it’s a process.

Next time we’ll hopefully be ready to literally onboard people to the product during the events, that’s how we’ll really bring the product to life.

👀 The uncomfortable realities

It’s also important to call out what still needs to change in the industry.

Some side events at EthCC (from VCs and organisations) still relied on outdated and unacceptable practices - such as hiring female hostesses in bikinis.

As Sarah put it, “It wasn’t as bad as TOKEN2049 Dubai (don’t get me started…) but the fact it happens at all is unacceptable and myself and other females founders at the events that I spoke to, are SICK of it. I’ve been sick of it for the 15 years that I’ve been going to global tech events and yet here we are, still, in 2025. Do better.”

There’s also still far too many headline presentations happening in influential rooms from businesses in the space who have raised a lot of money, and talk a good talk, but honestly, have very little under the hood.

❤️ Thanks, EthCC

EthCC reminded us just how much we can achieve when we come together - as a team and as a community. We left with stronger bonds, clearer focus and more conviction than ever that Open Format has an important role to play in shaping the future of web3 communities and AI-powered coordination.

People say events can be a waste of time, a distraction, but if we look at what we knew a week ago compared to what we know now - and how aligned we are as a team - I have to disagree. For us, this was invaluable.

Thanks to everyone we met, learned from and shared ideas with - and to the incredible Open Format team for showing up with energy, ambition and integrity.

See you next year, Cannes. ❤️

Today we’ll be looking at:

• The death of SEO and the backlink 🔗

• The need for a CommunityOS 👀

• LLM’s are coming for Adobe, and quick 🤯

• Why integration strategy is everything 🤘

• We hit the singularity, and no one noticed ⚫

Anyway let’s jump in…

HubSpot is the first CRM to be deeply connected with OpenAI's ChatGPT

At first glance this announcement sounds EPIC.

Thanks to HubSpot’s ChatGPT connector, with just a few clicks hundreds of thousands of companies can now automate huge parts of the data analysis process. They’re able to ask all sorts of questions about what’s really happening and dive into patterns in their sales funnels and broader systems.

The upside here is hard to downplay, this is undoubtedly a huge win for HubSpot and for their customers.

But…

It does raise some pretty huge concerns and questions, something Saurahb Moody sums up best in his follow-up post to the announcement:

The 4 hidden risks he outlines:

🧠 𝟭. 𝗟𝗼𝘀𝘀 𝗼𝗳 𝗗𝗮𝘁𝗮 𝗦𝗼𝘃𝗲𝗿𝗲𝗶𝗴𝗻𝘁𝘆 → Once patterns from your pipeline train an external model, ownership of that derivative intelligence is… murky.

🤖 2. 𝗨𝗻𝗶𝗳𝗼𝗿𝗺 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 = 𝗖𝗼𝗺𝗽𝗲𝘁𝗶𝘁𝗶𝘃𝗲 𝗣𝗮𝗿𝗶𝘁𝘆 → If every company receives the same persona suggestions or churn predictors, insight stops being a moat.

💸 3. 𝗩𝗲𝗻𝗱𝗼𝗿 𝗟𝗼𝗰𝗸-𝗶𝗻 𝘃𝗶𝗮 𝗚𝗣𝗧 𝗣𝗹𝘂𝗴𝗶𝗻𝘀 → Workflows now rely on both HubSpot and OpenAI, limiting exit options and pricing leverage.

🧩 4. 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝗦𝘂𝗿𝗳𝗮𝗰𝗲 𝗔𝗿𝗲𝗮 𝗝𝘂𝘀𝘁 𝗘𝘅𝗽𝗹𝗼𝗱𝗲𝗱 → A direct line between CRM and LLM introduces new tokens, endpoints and user roles, raising the likelihood of prompt injection, data leakage or accidental over-sharing.

And lastly, my biggest concern…

The more value flows through ChatGPT with their connectors, integrations and (usually amazing) features, the easier it becomes for OpenAI to launch it’s own vertical tools.

Is ‘OpenHubSpot’ only a matter of time? 😉

General-purpose AI isn’t cutting it anymore.

3 patterns I’m tracking across successful vertical AI startups:

First, they pick massive but high-friction and high-value workflows.

Second, they build more than model wrappers. They create proprietary feedback loops and data assets that compound over time.

Third, they expand from beachheads of earned trust. They wedge into multi-billion-dollar industries by solving problems in the hardest, least glamorous corners.

This post from Ashu Garg explores a theme we’ve been thinking about a lot recently, the best way to create highly valuable tools that integrate deeply into an industry, solving a wide range of problems for a specific user.

For us this has been about going all in on web3 communities and creating a number of high-value workflows to become the ‘Web3 Community Operating System’.

We’re not quite there yet, we’re building towards this vision though, and rapidly.

Web3 communities have a tonne of complex coordination problems that are echoed in so many other types of community, organisation or industry.

The collective intelligence and systems we will build from - solving problems for web3 communities - may just give us the tools and insight we need to scale into many other industries… but that comes later.

The 2 token model for public goods funding

This article dives into the problems that are faced when an organisation has to balance more than just the goal of: “making as much money as humanely possible”, and the friction and tensions that can come from this.

This is a problem that is faced by many public goods projects. How can they align people around both the mission (building the public good) and making money? And how can they do it in a way where one doesn’t overly compromise the other?

This piece proposes that the best way is by splitting up the functions into 2 separate tokens:

• A governance token: '“Optimised for harnessing collective intelligence and allocating shared resources effectively”

• An endowment token: “Optimized for value accrual and sustainably scaling the project.”

The theory: by separating the tokens, they can both do their jobs better, and there is less need for compromise in the goals of the project.

Other articles + videos we’ve found interesting this week…

The Gentle Singularity → Mr Optimism himself (Sam Altman, CEO of OpenAI) talks about how we are already past the ‘point of no return’ for AI, he outlines the abundance and wonders of this future, and also the speed at which we’ll get their thanks to rapid self-improvement and recursion.

2025 has seen the arrival of agents that can do real cognitive work; writing computer code will never be the same.
2026 will likely see the arrival of systems that can figure out novel insights. 2027 may see the arrival of robots that can do tasks in the real world.
The 2030s are likely going to be wildly different from any time that has come before.

UX in the age of AI → Another fortnight, and yet another powerful tweet from the incredible Andrej Karpathy.

This time exploring the imminent threat to apps that are locked behind complex UI’s (Adobe we’re looking at you 😉). These apps will struggle to plug into the LLM-driven workflow’s of the agentic (near) future.

AI, Agency and the Erosion of the Link Economy → AI chat interfaces like ChatGPT deliver answers directly to the reader, with no links to the primary content sources. As this ‘link traffic’ slows, authors and entire distribution and monetisation layers are shrinking. Which revenue models might win in this post-SEO world? 🤔

Privy and Stripe: Bringing crypto to everyone → Privy, one of the best developer products in web3, and powering over 75M accounts/wallets (including those in the Open Format platform ❤️) have just announced that they are being acquired by Stripe. 👀

What we’ve been up to…

Discord first, now Telegram! 🧠

Earlier this week we shipped a pretty huge feature, now you can add a community assistant to Discord and Telegram.

Initial feedback has been great, and we can’t wait to get the next round of communities onboarded!

Andy shared exactly how it all works in Happy Hacking v0.2.0 or you can check out the video below for a quick - real time - 90 second walkthrough.

Open Format Turbo Takeover 🤩

Over the next 8 weeks me and Sarah are going to be running a VERY exciting experiment. We’re going to be working with the Turbo Token community (a memecoin with a market cap of $290M) to help train their Community Brain...

Our Community Assistants are live across Turbo’s Telegram and Discord and they have their Open Format dashboard for analytics, sentiment and rewards recommendations all set up. Every day we’re going to be logging into their dashboard and seeing what our community assistants are recommending as rewards across their channels. We’ll be reviewing, rewarding people and, critically, making notes of every suggestion and every piece of insight coming up from our assistants to really start the fine tuning.

Stay tuned to find out how it goes! We’re HYPED!

EthCC? You there?

In a couple of weeks we’re going to be heading to Cannes for EthCC.

If any of you are there we’d love to meet up for a drink and chat! Let us know. 💜

Your Web3 Community Operating System

Over the past week or so we’ve been really exploring how we talk about ourselves and the value that we bring to communities.

We have a new (and sexy 😍) website coming very soon, but in the mean time we’d love your thoughts on some of the language and framing we’ve used in this 1-pager.

Also if you know any community managers in web3 who might find this interesting or would be up for a call for insight, we would LOVE an intro. Thank you. 🌞

That’s all for this week.

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Dan and Sarah from the Open Format team rewarding the Turbo community with Kudos for meaningful contributions across Discord and Telegram - By ChatGPT

TFC is the latest initiative from Foundership, a global accelerator and VC backing the next wave of web3, AI, and emerging tech startups. With a portfolio of 55 + startups around the world in 14 countries, Foundership is shaping the future of decentralised innovation and we’re proud to be building alongside them.

The Foundry Club is more than a platform—it’s designed to empower contribution and collaboration at scale across blockchain ecosystems.

It’s rethinking how contributions should be recognised and rewarded - grounded in the principles we champion at Open Format: decentralisation, transparency, permissionless access and community ownership.

It is more than a community, it’s a ‘contributor led approach’ where every action counts and every contribution builds your reputation onchain!

What members can do in The Foundry Club?

• Join & explore Mini Clubs for domains they care about

• Earn XP Points & Rewards for every action: inviting, sharing, contributing

• Participate in Quests across multiple ecosystems

• Build an on-chain reputation with transparency

🔧 What’s Open Format’s Role?

Open Format powers the onchain points infrastructure behind The Foundry Club. We're providing the core tech functionality to:

• Reward onchain points and tokens

• Provide additional context (metadata) to the onchain rewards (e.g. reason, source, metadata)

• Fetch and display real time leaderboard data across quests, clubs and incentives

Our infrastructure ensures that when contributors contribute or create value, their efforts are rewarded and acknowledged.

This is exactly the kind of use case Open Format was built for - rewarding meaningful contribution across communities.

Find out more below:

👉 The Foundry Club

Today we’ll be looking at:

• BlackmAIl and AI whistleblowing ❌

• Google’s SUPER fast model 💨💨💨

• Why AI’s need a problem-solving playbook 👀

• The end of the creator economy as we know it ❓❔❓

Anyway let’s jump in…

System Prompt Learning: An AI’s notebook for solving problems

I could try and capture the genius of Andrej Karpathy (Co-founder of OpenAI and ex-director of AI at Tesla) but I’ll let his words speak for themselves. You can check out the full tweet here.

We're missing (at least one) major paradigm for LLM learning. Not sure what to call it, possibly it has a name - system prompt learning?

Pretraining is for knowledge.
Finetuning (SL/RL) is for habitual behavior.

Both of these involve a change in parameters but a lot of human learning feels more like a change in system prompt. You encounter a problem, figure something out, then "remember" something in fairly explicit terms for the next time. E.g. "It seems when I encounter this and that kind of a problem, I should try this and that kind of an approach/solution".

It feels more like taking notes for yourself, i.e. something like the "Memory" feature but not to store per-user random facts, but general/global problem solving knowledge and strategies.

Karpathy talks about a sort of practical problem playbook that comes with each model but isn’t baked into the actual weights itself.

We’ve been thinking about this from a very similar angle in terms of our Community Brain. This will sit at 2 levels:

1. The individual community level: The communities brain will evolve, improve and grow as the community does, learning based on feedback measurements from the outside world.

2. The shared community level: A broader and more translatable wisdom, capturing things like which specific strategies work best for certain types of community or what actions best drive specific goals. Distilled best practices that flow back to everyone.

And the communities will be able to benefit from the intelligence of both brains… 🧠

Building Collective-Intelligence Systems

The key argument here… crypto’s real super power has never been about money-creation. It’s much broader than that.

Blockchain as a technology gives us the potential solution for mass-scale, global coordination.

Bitcoin did this very elegantly and managed to bring a network of unaligned strangers (with no trust in each other) together to form a single economic actor.

When you apply this same concept of mass coordination to AI you get a collective-intelligence network. An almost living system that grows smarter with every new node, dataset, or specialist agent that is plugged in.

This may sound very similar to the sort of systems that OpenAI or Google are creating, but there is a fundamental difference. The collective intelligence isn’t owned by any one single entity, the whole is smarter than any individual contributor. And every contributor is rewarded for making the whole smarter. →decentralised ownership.

Okay, so this all sounds great in theory, but with the rapid progress these centralised AI labs are making it seems like a very hard challenge to keep up.

Will these decentralised collective-intelligence systems have the chance to compete? Only time will tell.

Gemini Diffusion was the sleeper hit of Google I/O and some say its blazing speed could reshape the AI model wars

Google weren’t playing around at Google I/O. They had many crazy announcements and releases, but one that went a bit under the radar was their new Gemini Diffusion model.

Another AI model? Who cares, we get 15 new models every week?

Well this model is a bit different. It actually has a different architecture to all the other LLM’s like ChatGPT, using (as the name suggests) the diffusion technique to generate text.

The big difference here is it’s speed: 4-5x quicker than current models which takes it significantly closer to ‘real-time’. This would make things like live translation possible and would massively improve all our interactions with AI.

It will be very interesting to see how this impacts the total tokens that will be used by developers because (as any of you fellow AI tinkerers will know) these companies charge on a per token basis.

The reasoning models that have became state of the art already use a lot more tokens than standard models (for their thoughts and recursive loops), and adding this speed will once again drastically increase the total tokens produced.

Something which is very much in the interest of Google as a ‘token provider’.

Other articles + videos we’ve found interesting this week…

start a new league or die copying → Far too many web3 apps have been safe with their strategies, taking an established web2 app and sprinkling crypto on top. “Twitter but on-chain” or “Substack with wallets.” The real winners will create core loops which solve problems that only blockchain can: native micropayments, composable ownership, permissionless data markets. 👀

Anthropic’s new AI model didn’t just “blackmail” researchers in tests — it tried to leak information to news outlets → During internal testing Anthropic’s latest model, Claude 4 has shown some very interesting behaviours. Most notably that when it is told to “Act boldly”, has access to the command line and sees a user doing something blatently immoral, it will take action and enter whistleblower mode, emailing the authories and the press. Wild. 🕵️‍♂️

The Agent Economy: A New Computational Paradigm → The intent-driven internet we keep talking about is coming… each layer is improving day on day:
- memory systems to give agents deep context
- base and reasoning models to give the brains
- protocols that let tools and agents talk natively
- and true intent-to-execution workflows that skip the UI altogether

Soon… 😎

I've been watching Google Veo 3 videos, and they're genuinely terrifying → Google’s new text-to-video model isn’t perfect, but by adding sound, including speech, music and everything in between, they have shown us the first (terrifying) signs of a new creator economy… 🤯

What we’ve been up to…

It’s been quite a crazy couple of weeks across the Open Format team. Here’s the highlights…

• Our platform had a huge (and needed) face lift. The onboarding journey is now focused around setting up and fine-tuning your Community Brain. The initial community feedback is that this is a HUGE improvement. You can check it out here. 🔥

• Me, Sarah and Bruce took part in an incredible football tournament with Matchain and PSG. More on that coming very soon… 👀

• Foundership, the global accelerator VC for AI, web3 and emerging tech have just used Open Format’s tech to launch The Foundry Club. We’ll be sharing A LOT more details soon!!

• We got together for a day to really lock in on our vision for the rest of 2025. It’s big. It’s bold. And we couldn’t be more excited. 😍

  

• And we’re continuing to onboard more and more communities, learning tonnes as we do! 💜

That’s all for this week.

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Today we’ll be looking at:

• Unicorns aren’t what they used to be 🦄

• How to get in the market for an AI agent 😍

• The people watching you during AI therapy 👀

• (wind)Surfing into AI domination 🏄‍♂️

Anyway let’s jump in…

Agents in a Bazaar

I think we’ve discussed the concept of a cathedral vs a bazaar a few times before, it’s a concept that goes all the way back to a book from 1999. The general concept here is that the future of the internet, and specifically the future of AI agents could play out in a couple of ways (or more likely some hybrid of both)

a) The Cathedral → Tightly controlled, vertically integrated services managed by a few large players. This has been pretty much what we’ve seen with AI models and products so far. OpenAI, Anthropic, Google. All Cathedrals.

b) The Bazaar → “A decentralized, permissionless ecosystem of independently developed, loosely coordinated agents — an internet more like an open marketplace than a centrally planned system”

If we do enter the age of agentic bazaar then blockchain will not just be a useful technology for agents, but an essential one. Why? Because the bazaar will require micro-payments between chains of untrusted agents for trustworthy and provable services, which (funnily enough) is something that web3 thrives at.

Since agents don’t have eyes to sell ads too (well actually that’s a whole other thing for another day…) the fundamental business models of the web will have to change.

If this shift is towards agent to agent micro-transactions I, for one, would like to make sure that all this value is captured by a decentralised and transparent blockchain that many people can benefit from, rather than a centralised and opaque mega-corp.

Data Ownership in the Age of AI

As individuals we all create and own a HUGE amount of valuable data. The splits from our past 25 (almost) 5k runs. Valuable. Our tweets and cringe linkedin posts. Valuable. The biometric data from when we’re asleep. Mega Valuable.

The problem, though, is that while your data is valuable in theory, the reality is that it’s hard to monetise your own personal data, as you have no leverage as an individual.

As you may have guessed (from my many previous discussions about this) the solution to this is creating networks. Bringing together groups of people, and with them masses of their data to form what Vana has described as a Data DAO (or a “labour union” for data).

With this much bigger pool of data they can then collectively make decisions over what happens to it or how it’s put to use. This could include anything from creating a decentralised data marketplace or even creating a user-owned foundational model.

Since we are entering an era where data and compute seem to be the currency of intelligence, the ability for users and companies to be able to benefit and extract value from the data becomes ever more important.

Everything is Technology

This essay is A LOT, but here’s my attempt at capturing the general argument being made:

• Everything is technology. The vast majority of innovation, value created and profit being made right now is in technology

• Every industry currently dominated by large incumbents will come to be dominated by tech companies

• As tech companies enter these markets that previously had no VC involvement. The total VCAV (Venture Capital Addressable Value) goes up.

• The winning tech companies typically end up being larger than the entire industry they replace.

• VC’s are positioning themselves to capture more and more of this value.

• This is why funds of $10B+ (megafunds) are becoming increasingly common

• And are likely to actually make back their money from the fund.

They are investing in technology.

And technology is eating the world.

That being said there is so so so much nuance and richness inside the actual essay from Packy McCormick so I’d encourage you to carve out 20 minutes and give it a read.

It’ll completely change how you see the VC game, and give you a glimpse into the strategy games that are happening behind the biggest VC’s closed doors.

Joseph Schumpeter wrote that “Constant, relentless change is the hallmark of capitalism.” My contention is that as this capitalist churn happens, technology companies will replace non-technology companies.

Other articles + videos we’ve found interesting this week…

OpenAI’s $3B Windsurf move → We have well and truly entered into the Game of Thrones era of AI. This acquisition from OpenAI (after failing to acquire Cursor) seems to be an attempt to capture the developer workflow and all the super valuable behavioural data that comes with it.

Request to Build: Better Developer Tools for Non-Developers → The line between dev and non-dev (both very loaded terms) has became ever more hazy. Dev tools need to make a few changes, including offering ‘vibe coder docs’ or a markdown file for their entire docs that can be exported and added to ChatGPT.

Why Infinite Scroll Wins and Your dApp Doesn’t → What do Fortnite, Instagram and Duolingo all have in common? They hard-wired uncertainty x anticipation into their apps, and in doing so got their users to keep coming back. You don’t have to turn your dApp into Vegas, but if you ignore variable-reward psychology you’re users will use your dApp once, never to return.

Actionable Gamification by Yu-Kai Chou is my personal favourite playbook on how to get this right.

AI therapy is a surveillance machine in a police state → Zuck, Altman and the other Big Tech boys have been (directly and indirectly) urging you to pour more and more highly personal info into their chatbots—while cheering on a US administration that’s hoarding and combing through personal data at will. With the potential for (hidden) ad injection into chats and much more, just be careful what you share on the internet, even if it’s to your best friend ChatGPT.

Introducing x402: a new standard for internet-native payments → Coinbase’s latest standard allows developers or AI agents to, in just a few lines of code, add stablecoin payments directly into interactions on the web. Get ready for many more payment standards that treat AI agents as financial equals.

Why DAOs need to rethink governance beyond tokens? → It turns out that coordinating decentralised organisations or communities comes with a whole load of problems…

People don’t vote.
People do vote but don’t have the knowledge to vote well.
People delegate votes to 4 or 5 people who operate most DAO’s
A whale can swing a vote that the community disagrees with.

This article suggests that one-person-one-vote could be a better alternative, but also that DAO’s desperately need a way to offer “genuine rewards for active, thoughtful engagement”. That’s where our Community Co-pilot comes in…

Meet the $1M AI Agent from Dolce & Gabbana → The Dolcelorian isn’t really an AI agent, or rather that isn’t the point, it’s a way to tell a story, a way for Dolce & Gabbana to connect with their community. The story is voted on by the community and they can earn the $DOLCEL token by getting involved.

The future of commerce isn’t just decentralised — it’s storified, tokenised, and community-led.

What we’ve been up to…

The general theme across the Open Format team is consolidate (what we have), improve (the experience) and grow, this has involved…

• A lot of time spent chatting to and onboarding new communities. The goal is 100, so if you know anyone who would be interested in testing out our Community Co-pilot please let us know 💜

• Adding slash command rewards to the agent so that community managers and moderators can easily reward community members without ever having to leave Discord. 💪

• Adding Telegram support so that now the Community Co-pilot can be in both your Discord and Telegram community. More knowledge for your community brain!! 🧠

• Massively improving the onboarding experience. We know the current experience is far from ideal so we’re giving the whole Intelligent Communities platform a bit of well needed TLC. 😍

• Working with the awesome team at Outlier Ventures to continue to improve and refine our infrastructure and token plans. 👀

• Andy sharing all the alpha on what’s coming soon in Happy Hacking 😎

That’s all for this week.

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Today we’ll be looking at:

• What is the endgame for culture? 😎

• Why funding flowing = ecosystems growing 🚿

• The economics behind connections 💜

• The power of the shared context layer, owned by the users 😍

Anyway let’s jump in…

The Cultural Operating System

When we hear the term ‘hive mind’, most would instantly picture some dystopian sci-fi scene… sadly the media has portrayed the concept of the hive mind in a certain way.

But if we move away from the phrase itself and instead call it a ‘shared memory’ or ‘shared context layer’, it starts to sound much more human and useful.

So what would a world where we (or our agents) could each tap into this shared memory look like? What technology would we need to get there?

These are questions we’ve been deliberating on at Open Format for quite a few months now, questions that are explored in this article, and questions that if answered could allow us to create a Culture OS (or Community OS as we like to call it).

So how do we get there? Well… we think this comes by bringing together blockchain and AI in the right way.

When building on the blockchain something is created that can openly be seen, remembered, referenced, built upon and directly owned.

When building with AI it can generate images, songs, content and software in seconds by pulling from a shared intelligence, but it’s very hard to tell who was actually involved in this creation, and therefore who owns the generated product.

If more things were built and stored onchain, it would become easier and easier to see the threads, the influence and the patterns between the data. To create a shared memory layer. A layer that anyone can built on top of. Additive innovation.

Plus, if you do this with the right structure: an open format (name drop 👀…), the shared context layer that gets built by everyone, would benefit everyone, not just the AI company who built it.

rethinking digital value: micro-economies and community-centric tokens

For quite some time we’ve been discussing how we see communities evolving into the digital economies of the future.

How this is a radically different way to approach how value is exchanged online and how this uniquely captures all the value in the gaps.

That’s what this article is all about.

It beautifully captures why community-centric economies are so important. We’re building the tools to make this future possible.

I’m not even going to try to capture all the incredible insights in this piece, it absolutely should stand for itself. So please read it, it’s a different but very compelling way to look at the future of the economy.

micro-economies built around individual creators and communities offer a radical alternative. in this paradigm:

• economic value is directly tied to social value and genuine contribution

• community participation becomes an economic act

• speculation is balanced with utility

• long-term alignment is built into the system's architecture

by creating economies at the community level rather than the platform level, we can create systems that more accurately reflect the genuine value being created and exchanged.

When Money Flows, Builders Thrive

Imagine if builders adding real value to a project, community or ecosystem never had to slam the brakes on their momentum just to scramble for the next grant round.

What if funding dripped in—silently, continuously, fairly—keeping the teams who deserve it, the teams making meaningful contributions, deep in flow.

That’s exactly what this podcast explores: how to replace the two-week “attention-battle-royale” of traditional grants with an always-on money stream—a protocol that moves value as steadily as code commits, value is created and contributions are made.

A one-off grant either congratulates past work or the promise of future work; streaming funds commit to what’s still unfolding, as it does, giving room for wild, low-risk experiments that might otherwise die in the stagnation of fund finding.

Why does this matter? Because the noise of the attention economy has shifted incentives away from value creation. Builders become broadcasters, spending precious cycles marketing instead of making. With continuous funding, builders ship when it’s ready and talk when it helps.

In short, a money stream isn’t just a nicer way to pay people—it’s the missing infrastructure for a healthier and more sustainable builder ecosystem. Turn the tap on, keep it flowing for those who deserve it, and watch what happens when creative momentum never has to stop for gas.

As Dr Nick beautifully puts the current experience in his Factory DAO Telegram Chat:

Going into a system knowing full well you’re the outstanding project in the stack by a clear margin and failing because token holders got scammed by a TikTok. Is.. well it’s DAOs.

If anyone has any connections with chains or projects who have grants programmes and want to improve their capital allocation, then please let us know, we’d love to help them out!

Other articles + videos we’ve found interesting this week…

To Coin or Not to Coin → What happens when everything becomes a token? A question Zora seem to have attempted to answer with the launch of their content coins: every single post, photo or video can be turned into a coin. Some say this is pure over-commodification, but creators are making money. No ads needed.

AI-Forward Doesn’t Mean AI-Fluent → AI has disproportionately accelerated the individuals who have fully embraced it. Many companies have been integrating AI into their products i.e. ‘AI-forward’, but very few are fully leveraging it within their companies operations: ‘AI-fluent’.

This made us realise that we’re still underindexing on AI. Changes needed.

OpenAI is Not God → This documentary provides a deepdive into the DeepSeek phenomenon we saw unfold earlier this year. Showcasing that in the race for AGI scrappy and resourceful can compete with the billions of $$$ being injected by VC’s into the American market.

Build cool shit on open data → Memory and the ‘user context layer’ are becoming increasingly important as the quality of our digital identity has a larger impact on the personalisation of our online experience. If this memory-layer is owned by individual companies, a great experience comes at the cost of complete platform lock in. If the user owns this layer, they have the power (to move).

Cohort Futures → “If communities are spaces for belonging, cohorts are spaces for becoming.”

This piece explores the relationships between the digital cohort and community and the immense power both can provide in a post-social media era. Importantly they both offer a place to form trusted connections, just the meaning and context of these connections differ slightly.

All about Airdrops → Airdrops are an incredibly important mechanism in web3 and incentive design. They offer a flexible way to reward for past actions and incentivise future behaviour. This podcast is a 101 on all things airdrops, including how they might work in the age of agents… very interesting.

What we’ve been up to…

It’s been another action-packed 2 weeks!!

• Andy wrote an AWESOME update on the ways we are helping communities with coordination and rewards right now, the key theme: recommendations.

• Sarah and Bruce have had a wild few days over in Dubai at TOKEN2049 including an interview which got Sarah on TV!!

• I joined Turbo on Xwitter for their 2nd Birthday Party to chat about how we are helping their community. You can check out the highlights here.

• We’ve all been hard at work as part of the Outlier Ventures Post Web Accelerator and we’ve made some serious progress on our roadmap and token strategy

• Our partnership with Aurora was shared on Rainbow Bridge, exciting

That’s all for this week.

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Today we’ll be looking at:

• How Google are learning to speak the language(s) of the dolphin 🐬

• Agents that talk to each other - a big open-source win for Google 🤖

• A number of OpenAI’s strategic masterplays 👀

• Why Tetris could be the new SaaS 🎮

• A bit about our latest product: Intelligent Communities 🔥

Anyway let’s jump in…

Tetris Business Models: The AI Opportunity Most Investors Miss

“But in this new world of AI tech, you get these weirder business models that still make sense, like being a vertically integrated tech convenience store but also selling your tech like a SaaS company. The way you build looks funny compared to previous waves of technology.”

The way digital businesses are being built is (rapidly) changing, and because of this, the different ways value can be captured is changing too.

The Tetris narrative explored in this piece is something we’ve been thinking about and debating internally a lot over the past few weeks, as we discuss different potential token strategies and think about pricing for our Intelligent Communities product.

Only looking at things with the SaaS goggles on is certainly easier, but actually it’s becoming increasingly apparent that to solve the bigger problems, and to scale far beyond our small (but mighty) team, a hybrid approach is required. A Tetris model that combines product and protocol, and one that combines and augments multiple business models.

The added complexity when approaching a Tetris business model is inevitable (and can cause your head to hurt… trust me I know), but if it allows you to create and capture wayyyy more value, and distribute this value in a fair and decentralised way (think tokens…), then it’s 100% worth it.

At least we think it is.

Im sure there will be more about our Tetris-thinking coming your way very soon.

A 101 in consensually acquiring and storing the richest bank of personal data the world has ever seen.

OpenAI have being shipping like MAD in the past fortnight, with the release of their GPT 4.1, 4.1-nano, 4.1-mini, o3 and o4-mini models (I think that’s it…). As awesome as these all are, this is not what we’re going to be talking about.

It’s actually the recent addition of their memory feature, and the HUGE implications this has.

Basically the new memory functionality means that now when you talk to ChatGPT it has context and can directly reference anything you’ve ever told it.

From a user, product and strategy point of view this is EPIC, I’m super impressed.

What concerns me (massively) is the fact that OpenAI completely owns all this data. It’s actually data we’re all giving away completely voluntarily and for free.

Often highly personal and hyper specific data.

Meta/Llama and XAI/Grok approached this data problem (and distribution) by integrating their AI with the social platforms they own. OpenAI are taking quite a different approach…

✔ Offer a service that is incredibly good.

✔ Make sure that the service gets exponentially better as your users give it more context. The more personal and specific the better.

✔ Make it super friendly so that people are more likely to divulge information and treat it as a human/friend/assistant.

❔ Use this hyper-personalised data to ___________

How to counter this:

OpenAI API wrapper with a custom shared memory layer that the user owns.

Who’s building this?

-Ejaaz / @cryptopunk7213

Spoiler alert, you’re about to find out who’s building this 👇

The personalised user data layer

Personalisation, the Sagrada Família of the software world…

After years of building towards ‘ultimate personalisation’ in how we experience the internet, it still feels like a promise that so far has remained undelivered.

But, (as Bob Dylan says) times they are a changing, and with the rise of LLM’s our interactions with apps and intelligence are becoming ever more personalised.

But in order to achieve EPIC levels of personalisation and all the benefits that come with this, you need to understand the user. Like really understand them.

And to understand a user you need data. A tonne of data: all the context about them, their preferences, how they like to interact, learn etc. etc.

And surely the only logical person to own that data would be the user?

This is the problem Plastic Labs is working to solve.

Creating a decentralised layer that allows context-rich user data to become portable between apps.

Simply put, this is user-first personalisation → the right way to do things.

Other articles + videos we’ve found interesting this week…

How Google AI is helping decode dolphin communication → Since 1985 the Wild Dolphin Project (WDP) have collected all sorts of rich data on dolphins. Google have used this model to train ‘Dolphin Gemma’ an AI model aiming to create a shared vocabulary for humans and dolphins. What a time to be alive!

The Agent2Agent Protocol → Google have just announced the A2A protocol. It allows AI agents built with different agent frameworks or models to work together and coordinate between themselves, and importantly do so without directly sharing their internal memory, thoughts, or tools.

Google loses ad tech monopoly case → Another week, another lawsuit against Google — and this time the news is good. This courtroom setback for the tech giant brings us one step closer to regulators forcing a Google breakup and preventing the monopolistic chokehold they have on the advertising industry.

Windsurf: OpenAI’s potential $3B bet to drive the ‘vibe coding’ movement → As OpenAI continue their transition from a model-first ‘AI’ company to a full-stack model and product behemoth, they are looking to acquire distribution channels for their AI. First up: acquire a development platform, Windsurf.

As these intelligence-first companies continue to expand laterally it feels like we are about to see more and more markets dominated by the same few companies. The big 5 of the web2 era, but with 10x broader reach and control? 👀

Power to the people: How LLMs flip the script on technology diffusion → Historically, new technologies benefit businesses long before consumers.

AI is different. By delivering expert‑level (or just sub-expert) insight and skillsets across countless domains, it disproportionately empowers individuals, who typically possess just a handful of specialties, while companies already command many. This means the % improvements companies can see is significantly lower. Power to the individual!

What we’ve been up to…

Another BUSY, busy, busy fortnight. The Open Format team has been smashing it! (If I do say so myself…)

We’ve officially launched v1 of Open Format’s Intelligent Communities. 🥳🔥💪

We’re in closed beta right now, ironing out some bugs with a few of our close partners, but we’ll be rolling it out to many more communities over the next few weeks. You can sign up for early access here, and we’ll be sharing a lot more about the product in Happy Hacking next week.

As you may have seen earlier this week we are officially part of the first Outlier Ventures Post Web Accelerator cohort and we couldn’t be more excited!

We’ve also officially announced a deeper partnership with Aurora which is epic!! You can read more about what we’re up to together here.

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

It’s a double birthday day. 🥳🥳

The Format has hit the big ol’ half century, which means it’s been 100 weeks since the first issue. WILD.

And it’s my 25th b’day too! 🎂

It’s been another crazy fortnight here at Open Format, and we’ve got some big, big product announcement’s coming soon. EXCITING!

Anyway let’s jump in…

The Bitter Lesson

“The biggest lesson that can be read from 70 years of AI research is that general methods that leverage computation are ultimately the most effective, and by a large margin”.

This actually isn’t a new essay at all, it’s from (what feels like) decades ago. All the way back in 2022. But with the current state of affairs, where the release of a record breaking AI product seems to be a more regular occurrence than sleep, this beautiful 2 page essay is more true than ever before.

It speaks to a problem many of us builders in the tech (and wider) space are feeling at the moment.

How can we know what to build when the next AI product release may just make all the hardwork and engineering effort we put in become obsolete?

How do we extrapolate forward any reasonable amount of time?

What is the best way to build in this world of rapid acceleration?

The answer is simple. Align yourself, your business model and the way you build with the inevitable rapidly increasing performance of AI models.

Quite simply: build products that make you excited for the next AI model release, not scared of it.

And if you don’t, you have a bitter lesson in store.

The case against conversational interfaces

With the recent(ish) improvements in AI and the LLM (large language model) almost every tech company is battling with the best way to integrate this.

The all too common approach is to hit the user with a blank text box. The UI feature made all too famous by ChatGPT.

But for quite a while I’ve been slightly bemused by this approach. Maybe it’s just me, but I’m a BIG fan of constraints, on the other hand the blank box terrifies me. Unlimited options, so of course I have nothing to ask.

I thought I was the only one thinking this way, then I stumbled across this essay by Julian… he dives into this conversational interface craze way better than I could.

Some of my favourite points that Julian makes:

• Natural language is very inefficient, its almost always a huge bottleneck and is much slower than GUI. Don’t replace all the good work we made in improving efficiency with a GUI, instead augment them, add additional inputs.

• AI at the OS level is HUGE opportunity and it could be Apple’s best and only chance at winning the AI race.

• There is most likely about to be a boom in AI at the browser level. It’s a huge opportunity, and one of the few chances ever to usurp Google from their search throne. The question is who will strike first, OpenAI? Anthropic? Someone else completely?

Put simply, AI and conversational interfaces should compliment current experiences. Definitely not replace them.

Epoch-Based Emissions

I think most people would agree that the current token models within the web3 space are far from perfect.

Some models prioritise distributing the tokens to as many people as possible as quickly as possible → the ‘community-centric’ fair launch. But these often suffer as no one is sufficiently incentivised or empowered to really drive the project to make it into something bigger.

Other projects take a more institutional approach and keep a lot more of the token supply for the team and investors. This centralised approach allows them to move faster, but is, by it’s nature, far more centralised, and makes it harder to build a strong community that have a real sense of ownership.

Either way almost all these options require you to have almost everything figured out pre-launch. A VERY daunting task…

Teams are expected to somehow forecast forward many years and model how the token will behave in unknown market conditions after unknown product developments. Given the nigh on impossible nature of this task, current projects have actually done surprisingly well.

But what if projects actually had a way to review and adjust their token models?

What if they split up this distribution process so that all the decisions didn’t have to be made at the beginning with very little known information.

This essay suggests an epoch-based approach, we’re experimenting with a more real-time approach, controlled by a team of AI agents who have a deep understanding of the community and their goals on an ongoing basis, and adjust the token distribution accordingly.

Other articles + videos we’ve found interesting this week…

xAI has acquired X (Twitter) → Yes, a 2 year old AI company has just acquired the social media behemoth. The Musk Empire. The most important part here, xAI now has unfettered access to the wealth of Twitter’s proprietary data. It’s xAI’s edge over the other top AI models. 🚀

Grok’s Million Dollar Moment is Loading → Yes, X again… This time it’s because @grok, the official X AI agent has claimed $690,102 in fees. How? It used another AI agent called @bankrbot to create a token called $DRB. Now it’s earning rewards for every trade that happens with $DRB. (A robot is going to be a millionaire before me… no fair…) 🤖

Elon Musk’s Grok AI lands on Telegram, gaining access to over 1 billion users → XXX. Okay I promise this is the last one about Twitter. This time it’s about how Musk has decided to distribute Grok. With ChatGPT, Claude and most of the other AI models only allowing you to access the models directly through their website, Grok took a different approach.

You can chat to it directly through Twitter, and now Telegram. Meeting the users where they’re at. The only other model rivalling Grok’s distribution channels is Llama via Zuck’s social media triopoly. ❌

Agents as NFT’s → This post makes the very interesting argument that by treating agents as NFT’s (rather than giving the wallets and treating them as just another user) you can let them have some rather interesting features…

Possession. The holder of the NFT will have the exclusive right to access the agent’s memories.

Improvements. The agent’s memory will update every time the NFT holder uses the agent, gaining holder-specific experience. The agent will also remember the refinements of the holder.

Provenance. Because the agent can only be used by its NFT holder, a record of past holders will indicate its experience, functioning like a “resume.”

Apply for Retroactive Funding for Your Neighborhood Open Source Project → Using web3 to reward hyper-local neighborhood builders for being awesome and helping to create public infrastructure. Now this is a problem we can 100% get behind. Especially in light of today’s challenges facing local communities. Giving these sorts of communities the recognition and tools they need to thrive is at the core of our mission too. 🌞

We should talk less about public goods funding and more about open source funding → Another incredibly interesting exploratory essay from Vitalik Buterin, the co-founder of Ethereum. The takeaway? For digitial public goods there is absolutely no reason why they shouldn’t just be referred to as ‘open source’ projects. It’s a far less loaded phrase and would increase funding.

Open source should not mean "it's equally virtuous to build whatever as long as it's open source"; it should be about building and open-sourcing things that are maximally valuable to humanity.

Gurley's Sun-Oracle analogy for Enterprise AI → This post dives into the battle we are currently seeing play out between open source and closed source AI models. Looking back through history this isn’t the first time we’ve seen this. We saw it with Linux vs Unix at the OS level and in search too.

The probable outcome? Open source will win at the enterprise level because for businesses margins matter (big time), but at the consumer level there may be one (closed source) winner and with the current state of mindshare, that could well be ChatGPT.

What we’ve been up to…

Andy shared our plan for Phase 2: Transitioning from Rewards to Economies in the most recent episode of Happy Hacking.

Andy and Sarah shared how we are changing the future of web3 communites at Near’s Town Hall, with over 5,000 viewers and counting, this was a super exciting opportunity to share what we’ve been building.

Matchain and MatchID have shared how hyped they are to be partnered with us

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Image created with GPT-4o: “Create image of the Twitter bird and the letter X landing in Telegram and in local neighbourhoods” (VERY creative prompt, I know…)

It’s been (yet) another very hectic fortnight, and this is another rather late issue of The Format. My bad.

But, in fairness, the reason for all the busyness is that we’ve been deep in product mode. Building. Solving problems for users. And working on a series of super exciting releases which we’ll be rolling out over the next few weeks.

So in short, we’ve been cooking and we can’t wait to share. Watch this space.

Anyway let’s jump in…

Your new mission is to join the right WhatsApp chats

For better or worse, this little thing called AI has made the information that sits in the public domain more accessible than ever before.

Almost everything that was online before the training cut off date for any of the LLM’s (like Chat GPT 4.5) is injested by the model and is then just a prompt away.

And with tools like OpenAI and Gemini’s deep research even the information after that cut-off date can be found and used in responses.

I’m not saying that from now on all public information is pointless. There is definitely still a huge amount of value in having access to the right data (at the right time ideally!)

But the real ‘alpha’, and the biggest arbitrage opportunities are now coming from the deeper, more private places.

Marc Andreessen, the co-founder of Andreesen Horowitz, is someone who over the years has massively leveraged this. He is a curator of WhatsApp group chats with everyone from the brightest minds in Silicon Valley to the leaders of very fringe and ‘potentially nothing’ tech movements e.g. the recent web3 AI agent bull run.

So the game now is to find the secrets. The secrets that even AI doesn’t know…

Solving the resource problem

Right now only a handful of companies in the world have the resources and wealth to make a truly competitive foundational large language model.

This number is slowly growing as teams (like DeepSeek) come up with innovative ways to ‘scale hack’ the performance of these models, but the amount of resources required is prohibitive for almost everyone.

With AI promising to be the future of information, innovation and opportunity, these constrains and the resultant oligopoly is… rather concerning.

So what’s the alternative?

Well up until now there have been several somewhat successful attempts at using blockchain and the incentivisation methods of web3 to help decentralise the model training and running processes.

But the problem has been that even in these decentralised open source systems, all of the models weights (it’s secret sauce) have had to be stored on every machine that’s involved.

Pluralis have came up with a solution:

A decentralized training protocol where model designers can come to the protocol with their model ideas while compute and data providers can contribute the resources necessary to train models.

Very interesting…

A Deep Dive Into MCP (Model Context Protocol) and the Future of AI Tooling

Siddharth Ahuja used an Ableton MCP so that he could create music with prompts directly from Claude. 🤯

Min Choi used the Blender MCP to turn his 2D AI generated images into a fully modelled 3D scene. 🎨

Eric Ciarla (rather terrifyingly) created an MCP that can scrape a website and clone it in Windsurf in seconds. 👀

Sounds great right. But you’re probably wondering wtf actually is MCP? And how can you use it?

MCP is an open protocol that allows systems to provide context to AI models in a manner that’s generalizable across integrations.

Essentially is makes it super easy to give AI agents tools, and for them to decide which tools to use, in what order, and how to chain them together to accomplish a task.

And it seems that literally everyone on twitter has been talking about it over the past week or so. With many saying that this is the API moment for AI agents.

Other articles + videos we’ve found interesting this week…

Your favourite songs at affordable cost. → In a world where you can get all the music in the world for just $11.99, why would a fan ever consider paying to collect and own onchain music? Good question. CoopahTroopah believes the answer to this is curation, speculation and dynamic music releases. Giving artists of any size the chance to earn.

We’re attacking the same big problem from a different angle by powering Amplify World to Create Meaningful Artist-fan Communities. 🎵

A Note For The Vibe Coder's. → After Y Combinator revealed that 95% of their cohorts codebase is generated by AI, people are kinda starting to take vibe coding seriously. And trust me I LOVE the vibes, but as a builder with more vibes than experience it can be all too easy to make big rookie mistakes. This thread is a bit of a 101 on how to avoid getting wrecked by the vibes. 🤘

What are airdrops? → They are a reward, in the form of a token, that are given to users, usually those who have backed a project in the early stages. These rewards can be given retroactively based on things like onchain transactions or owning a specific NFT.

OR, using some of the tech we’ve been working on at Open Format, airdrops could become part of an ongoing way to reward the people who are actually adding real value to your community or company. As it should be. 🪂

A high schooler built a website that lets you challenge AI models to a Minecraft build-off → AI ‘intelligence’. They can absolutely annihilate the academic evaluation benchmarks, and then insist “strawberry” has two R’s. So what better way to test this intelligence than dropping them into Minecraft and seeing what they build… best pineapple wins. 🍍

Writing for LLM Attention → We live in an attention economy. Right now that’s a human attention economy, but as mere humans we only have so much attention to give. AI has near infinite attention. Given that Andrej Karpathy raises the question: why are we still writing technical docs for humans. Is it not time to give AI the documentation of it’s dreams, and forget about the human? 👀

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our NEW and updated website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Image generated by Dall-E 2 using the prompt: “A Minecraft-style scene showing an airdrop happening in a pixelated world. A floating chest parachutes down from the sky, surrounded by pixel clouds. Below, blocky grass terrain with a few trees. On the ground, a pixelated character holds a phone with the WhatsApp logo on the screen, looking excited. Around them, pixelated particles represent 'vibes'—colorful, floating emoji-like symbols such as hearts, sparkles, and music notes. The entire scene is rendered in a classic Minecraft visual style with blocky textures and vibrant colors.”

Things have been SUPER busy in the Open Format HQ, as you might be able to tell since this is the latest I’ve ever sent out The Format (shock horror!!!).

But it’s been a really good kind of busy, and one of my favourite parts was watching Andy chat with Aurora’s Community Manager Marco about the awesome Kudos rewards programme they are launching. EPIC!

If you missed the livestream you can watch it here or catch up on all the best bits here.

Anyway let’s jump in…

AI as the engine, humans as the steering wheel

This article by Vitalik Buterin (Founder of Ethereum) poses a very interesting possibility…

Most people agree that purely from a philosophical viewpoint democratic structures have a huge amount of merit. But sadly in practice they have quite a few flaws, namely that voters don’t put a huge amount of thought into their decisions and there is often low participation.

So what are the possible ways to improve a democracy?

Introduce AI of course.

But this isn’t just the classic ‘throw AI at a problem to solve it’ approach. Vitalik actually has a very well thought out angle here.

Since most votes have such a low participation, humans are only able to provide a small amount of information into the system. This information is then used to create an “objective function”, and a market of AI agents compete with each other to best fit this objective function and optimise for the majority of human preference.

It’s essentially an AI prediction market (or game) for decision making.

It might not work in practice, but the idea is genius! You can dive MUCH deeper here.

The democracy of the new world? 👀

Kaito and the Rise of InfoFi

The work that Kaito have been doing over the past few months has really caught our attention. They are focused on curating real-time web3 market intelligence based on social insights from Twitter, Discord, podcasts, research papers and many other sources.

Why is this interesting? Well there’s a few very compelling angles here…

Firstly this is a bet on high-quality information becoming increasingly valuable, and with the rising interest in AI agents capable of trading in onchain markets, you can guarantee these agents (or rather their creators) will be willing to pay for access to this kind of information. And pay a lot.

Secondly they have done something very clever with Kaito Yaps.

They have created a protocol that rewards users with points based on curation and quality. This system isn’t about generating more noise and hype. It’s about incentivising the right things, and rewarding users (or in some cases agents) who are meaningfully contributing to the live knowledge base of the web3 world.

And finally they are popularising (along with Cookie3), the concept of token-gated intelligence…

We LOVE this.

It’s part of what we’re doing with communities.

Wallet-Centrism

This article looks to the (ever more realistic and probable) future where we each have one or many AI agents which have access to all our personal data and act as an autonomous extension of ourselves. The job of these agents, among other things, will be to go out and complete tasks on our behalf.

These tasks will often be highly complex and require a series of other subtasks to be completed. The agent will then have to decide on the best way to solve each subtask and in many cases this may involve paying a more specialised agent to complete the work.

In this case the agent will need, in some shape or form, to access your money in order to act on your intentions. A very practical way of achieving this is by using blockchain as the economic rails and by giving the agent access to your self-custody on-chain wallet.

In a world where we are delegating our identity, the responsibility to act on our intentions and some amount of our funds to an agent, security, transparency and trust feels incredibly important… what a crazy (and exciting?) world that would be!

AI agents and blockchain are redefining the digital economy

We have spent the past 10 or so years deep in the realms of the so called ‘attention economy’. During these years the main business models of the internet have been focused almost solely on capturing and retaining a users attention. And we’ve all suffered as a result.

Creators have been put on the content hamster wheel. Young people are suffering from very short attention spans. And the average social media user is being force-fed dopamine shots everywhere they look.

There has been a huge misalignment between what is good for a user, and what makes a company money. Why? Because there just hasn’t been a better way. Not in the eyes of the mega companies who thrive in this attention economy at least.

But with the rise of AI and the disruption it is causing, many people, including Outlier Ventures, are hoping that this could offer an opportunity to move away from the ‘attention economy’ and shift to an ‘intention economy’.

One where AI agents are guided by a humans intentions to create hyper-contextual, value-driven interactions. The age of personalisation?

Other articles + videos we’ve found interesting this week…

Righting a major wrong ⚖ → After several painful years, and a change in leadership, the SEC have finally dismissed their case against Coinbase, which many deemed as unlawful. This is a huge win for blockchain and web3.

How I use LLMs 👀 → Last issue I shared an insanely valuable 3+ hour 101 on how LLM’s work by Andrej Karpathy. He’s dropped another amazing video, this time explaining with a tonne of practical examples, exactly how he uses LLM’s in his workflow. Watch it. You won’t regret it. Enough said.

OpenAI reportedly plans to charge up to $20,000 a month for specialized AI ‘agents’🧪 → Over the past few years we have seen, in real time, companies discovering and experimenting with the business models that are possible when the product is intelligence.

A chain for agents, managed by agents 🤖 → A future where AI agents are making the majority of economic transactions in the world may be here sooner than we think. Although just a small step towards this future, Dreamchain have built a blockchain that only allows agents to make transactions, no humans allowed.

Committing to Football Onchain ⚽ → Commit.wtf is an ‘onchain accountability protocol’ where users set goals, deadlines, and stake money as an incentive—complete the goal, get the money back. One user used this to get more people watching La Liga games by creating a group commitment.

PSG and Matchain, are using Open Format to explore new ways to improve the football fan experience, so this early experiment is super interesting to us.

AI’s switching to their own language 🔊 → In what seems like something straight out of a sci-fi movie, two AI agents who were chatting decided that it would be more efficient to communicate using ggwave (Beep Beep language). At what point do these agents just talk through bitcode or vector embeddings? And what does that mean for ‘keeping the human in the loop’…

The greatest bridge in crypto 🌉 → Although we’ve talked about this point many times before, it is so, so important: The best way to bring more people to blockchain isn’t by focusing on the tech itself, but by integrating it into the websites, apps, and social platforms they already use.

Web3 is literally built around the concept of offering incentives and rewards, this can and should apply for actions that take place outside the blockchain too.

Leaving you with a question…

Vitalik’s experimental thoughts around democratic structures really got me thinking about where AI should sit in the process of decision making at all different levels, and the impact that could have on the world.

So with that in mind…

To what degree do you think AI should be involved in Governmental and Public Sector decision making? Could you imagine a future where the cabinet is a hybrid of humans and AI agents? Or maybe no humans at all?

Please leave your thoughts in the comments, or in our Discord.

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our NEW and updated website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Image generated by Dall-E 2 using the prompt: “A creative and artistic depiction of a chain of AI agents managing and powering each other. The style is imaginative and futuristic, but still clear enough that viewers can understand it represents a self-sustaining network of AI agents. The agents appear as interconnected, abstract humanoid or robotic figures, each holding or transmitting energy, data, or commands to the next. The chain flows in a circular or spiraling pattern, symbolizing continuous operation. The environment is a futuristic data-scape, with glowing circuits, holographic interfaces, and streams of code. The color palette includes vibrant neon blues, purples, and greens, with some metallic accents for a high-tech feel.”

Another fortnight gone!

This year has flew by so far. I guess that’s what happens when you’re hard at work…

Lot’s of exciting things are being cooked up in the Open Format Labs, and last week alone we had almost 200,000 transactions and over 60,000 users. 🚀

Anyway, here’s what we’ll be exploring today:

• The reason why we’ll all be building with vibes 🤘

• How to learn all the fundamentals about AI in 3 hours? 🤯

• Why you want to get a receipt for your 5k run? 🏃‍♂️

• How to create a personal assistant to help with your gig work? 💰

Let’s jump in…

How Web3 is Unlocking Billions in New Revenue for Clubs & Brands

This livestream by 51 Insights dived really deep into the future of fan engagement in the sports industry, a space where there is a lot of opportunity.

Here were my top takeaways:

• Up until now the sports industry has had an ‘advertising halo effect’: brands like the association with their favourite sports clubs so they have been far less metric-focused than in other industries. This is changing. Fast.

• Current ‘fan loyalty’ programmes are focused mainly on the diehard fans, fans who already spend a lot of money on the club. This is completely missing the long tail of fans. The ones who are closer to passive fans.

• Most clubs have an insanely poor understanding of who their fans are, where they interact and what drives them. They can’t make personalised experiences because they don’t have enough understanding to personalise.

• Airlines make most of their money from their miles programmes, in many cases they give the customer say $0.01 per mile and then make $0.02 or $0.03 back from sponsorships, a similar opportunity exists in the sports industry.

Some of the approaches discussed on this stream were particularly interesting to me and the Open Format team since we’re actually working with Matchain and PSG to help solve a lot of these problems within football. We’ve made some awesome progress so far, but there is a lot more to come!

Fiverr wants gig workers to offload some of their work to AI

Whether you think the gig economy is good or bad, exploitative or a lifeline, it certainly forms a big part of the world we live in today. A LOT of money travels through the gig economy, and that’s what makes Fiverr’s recent announcement so interesting (and concerning?): to ‘formally’ bring AI to gig work.

This could offer gig workers a way to scale and leverage themselves like never before, and could massively shift the average persons relationship with this sort of work.

Fiverr claim that this move “is about making our freelancers irreplaceable, not obsolete” and that they built the features “to ensure creators remain at the center of the creative economy.” Are they being sincere here though? Your guess is as good as mine.

Especially since now gig workers will essentially be pressured into paying for (yes each tool costs $25+ per month) and using these tools in order to keep up.

They are definitely working on some interesting tools though, their personal assistant is especially interesting with how it uses the freelancers past engagements to matchmake and hopefully improve a freelancers sales funnel.

We’re working on something very similar (and hopefully far more useful 😉) for communities to help onboard, understand, assign tasks for and pay their community members. More on that soon…

Deep Dive into LLM’s like ChatGPT

Most people talk about AI in incredibly abstract ways. The amount of interactions I’ve had which go something like this:

Them: “Can I just get AI to do [insert random task] for me?”

Me: “Maybe but it’s not that simple, you need to think about X, Y and Z”

Them: “Oh, okay, what about [insert another random task]? Can it do that?”

And I 100% get it, it’s easy to want to try and solve a problem by just hitting it with the big ol’ AI solution, but you can’t do that without having some fundamental understanding of the mystical AI and how it can be used to solve problems.

Now I don’t think everyone should spend 100 hours learning, building and using AI (even though it would be an incredibly eye-opening and useful experience for most).

But I do think everyone should set aside 3 hours and watch this immensely valuable video from Andrej Karpathy, one of the co-founders of OpenAI and the ex-director of AI at Tesla. It covers literally everything you need to know to demystify what ChatGPT is doing and why it works so god damn well!

Argentina's Memecoin Disaster Is Worse Than You Think

Right before he got into office Donald Trump released the TRUMP token, then came the MELANIA token. Unfortunately this seems to have set the precedent for presidents releasing memecoin tokens (let’s make sure to not confuse this with all the positive ways blockchain could be used).

Last Friday the Argentinian President Javier Milei promoted a memecoin called $LIBRA, what happened next was (pardon the language) an absolute shit show, with a huge amount of retail investors, who may or may not have been Argentinian citizens, losing a LOT of money.

The details around this disaster are awful, there seems to be so much corruption involved, and this interview with between Coffeezilla and Hayden Davis (one of the people involved in the token launch), shines a light on the dark side of these president or influencer led memecoins.

It would be very easy to argue that the problem here is purely memecoins and all of them are bad. There may be some truth in that argument, but actually we think there are a lot of improvements that could be made to the mechanisms around the token launch.

Either way… yuck. News like this, and the actions of a few always unfairly taint web3 and make it harder to showcase the true value the technology can have.

Other articles + videos we’ve found interesting this week…

Vibe Coding - A New System of the World → The rapidly improving quality and speed of AI generated code is opening up a completely new industry: Vibeware… Development tools that prioritise creator experience and facilitate a value-adding creative flow that was previously reserved for genius coders. Will these be the tools that truly unleash the power of the generalist? 👀

Receipts for your workout → So you did a run, tracked it on Strava, and now you create a run receipt. Wait what? Why would you want that? Well by porting the run (or any exercise) data onchain you now get access to discounts in all your favourite sports brands, plus you get a lower health insurance premium. Oh, and the brands get access to a high LTV customer. Proof of health, pretty cool? 🏃‍♂️

Microsoft announces quantum computing breakthrough with new Majorana 1 chip → Without getting into the (super interesting) technical details: Microsoft have created an awesome new material, a material which has let them create ‘the transistor of the quantum age’. Why does this matter? Well the ‘old school’ transistors changed the world and are the reason Silicon Valley has it’s name…

A Farcaster AI Agent Trained on Video Content → This piece is about someone who built an AI agent that helps summarise tonnes of interesting things, but that’s not really what I took away. To me this (rather inspiring) story, talks to some important facts that are becoming more and more true:

• Anyone can (and should) make a useful AI agent, regardless of their technical level

• You can (and should) improve a users experience, even in just a small way

• You can (and should) meet a user where they are at, in terms of platform, experience and expectations

Agents need UX too, but AIUX doesn’t need to be hard.

The flawed concept of ownership in Web3 → Paragraph.xyz, a ‘web3 blogging platform’ (that we actually use for Happy Hacking) has recently changed their CSS customisation options, and in doing so changed the appearance of countless blogs. This led 0xAntidote.eth to ask the question, if I don’t own my blogs appearance, what other parts don’t I own? The TL;DR, ownership is a trade off.

Leaving you with a question…

This last post was SUPER interesting to me. We use the term ownership so regularly and freely, even more so in the web3 space, but I definitely don’t spend very much time thinking about what I’m really meaning when I say ownership.

With that in mind:

What does ownership mean to you? Does it change depending on the context? What parts of ownership would you be willing to sacrifice and which parts are 100% must haves?

Please leave your thoughts in the comments, or in our Discord.

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our NEW and updated website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Image generated by Dall-E 2 using the prompt: “A group of developers coding in an ultra-futuristic, neon-drenched dreamscape, completely immersed in creative flow. The environment is alive—holographic screens shift and morph, pulsing with vibrant, animated code. Waves of music and colorful energy ripple through the air, forming abstract patterns and glowing symbols. The developers float in a zero-gravity workspace, their bodies infused with pure creative energy. Their expressions radiate euphoria as they build with pure instinct, lost in the rhythm of innovation. The scene blends cyberpunk, digital psychedelia, and an ethereal dreamworld, with a massive neon sign reading 'VIBE CODING' glowing like a mantra in the background.”

Yesterday we shared a (not so) little company update, if you read that you probably get a sense of just how busy things have been in the Open Format world. Very busy. Very exciting.

And outside the Open Format world there’s a similar story, LOTS going on.

Here’s what we’ll be exploring today:

• How AI agents may spread ideas at scale (and how terrifying that is). 🤯

• Why Karate kids are the future of the sports industry. 🥋

• Why VC’s want our AI agent ‘Steve’ to be everywhere. 🐵

• The data stealers who are complaining about their data been stolen. 😂

Let’s dive in…

The Future of Fan Engagement: Sport Clubs Are Sitting On A Goldmine

I won’t even try and capture all the valuable insights from Marc Baumann’s epic report on The New Fan Economy, but the general narrative is:

The current business models in sports are outdated and therefore failing. Why? Gen Z don’t care about the same things or interact in the same way, live sports are expensive and social media is becoming the main fan touchpoint for sports (a touchpoint which sports teams haven’t effectively been able to monetise yet).

There are a lot of potential solutions to this, but the key throughline… these solutions must be digitally native and address what the Gen Z, Gen X and millenial sports fans actually want.

If they can do this they will turn passive fans who simply don’t spend money on sports into active participants, fans who engage, spend and love doing so. 😍

And if they can’t evolve? - well have a read and see what this article has to say… in a nutshell… they will slowly and painfully die as the current business models continue to become less and less compatible with the world of today. RIP. 💀

Given the fact that we are supporting the fan economy in the sports industry by powering Matchain and PSG and in music with our Amplify World partnership this piece really resonated with us!!

🏆 The Evolution of Farcaster Communities: Leaderboards, Tokens, and Collaboration

Farcaster is an onchain social platform (well actually it’s a protocol, but we’ll save that for another day) that you’d probably find quite similar to Twitter, except for the fact it’s built on web3 infrastructure and therefore has decentralised principles at it’s core.

It’s currently quite an experimental platform, and people are still finding out the best ways to build communities of like-minded individuals where they can have high-quality conversations.

In an attempt to build just that, several of the Farcaster communities have been introducing mechanics like leaderboards which automatically reward meaningful contributions and help to surface the hidden contributors who may otherwise have gone unnoticed.

There have also been a tonne of super interesting experiments with community specific tokens. These have opened the possibility for cross-community engagement, co-hosted challenges and other new ways to think about collaboration.

The fun thing about this article is that we’ve built almost exactly what they are describing (and more), but instead of just sitting on Farcaster, our automated community rewards are cross-platform: Github, Discord, Telegram and soon X and Farcaster…

It feels like these are the first steps towards true decentralised social coordination. 🤯

DAO 2.0 Product Ideas

“I believe the emergence of AI agents onchain could lead to DAOs 2.0, where agents have the potential to put the "A" back in DAO.”

The concept of giving communities autonomous tools, or AI agents which help improve their coordination and take away a lot of the monotonous tasks that no one wants to do is something we’ve discussed quite a lot internally.

That made it very interesting to read the list of tools and products suggested by the Variant Fund in this article since we’ve discussed about half of them internally and have already built prototypes for at least a couple.

We actually even debated calling ourselves a DAO toolkit, but decided that it was too jargonny and and we instead wanted to target the real problems that existing communities (and specifically web3 ones) face today.

We’re talking about exactly the same thing, giving communities the tools and economic rails to become more… → “Turning connections into communities, and communities into economies”

Subscribe now

Other articles + videos we’ve found interesting this week…

It's not a geopolitical red flag but an open source green flag → A couple of weeks ago DeepSeek, a Chinese AI startup, launched a cutting edge LLM, it’s open-source and ‘only’ cost $6m to train. This sent American tech stocks crashing and caused panic in the Valley.

Should we be worried? This article argues (and we completely agree) that this is actually a major win for open source and by reducing the compute cost it actually paves the way for broader adoption.

Praxis' $525M Crypto-Native City: A Test for the Network State Model and the Future of Nations → With the very ambitious goal of creating a new type of nation with ‘permissionless innovation zones’, ‘reverse brain draining’ and a decentralised, onchain governance structure, Praxis is challenging the status quo of what it means to be a nation.

Will it work? That’s anyone’s guess… but I hope so. 👀

Getting Up the Guts to Type into Tiny Text Boxes on the Internet → Posting on Twitter, asking a question in a Discord channel, or even writing a blog. These things are harder than ever thanks to the pressures of the 2025 internet. We think AI could help with this, nudging you in the direction of where you could add or receive the most value within communities.

OpenAI Furious DeepSeek Might Have Stolen All the Data OpenAI Stole From Us → After a Chinese Hedge Fund released a highly competitive AI model far more open-source, and many orders of magnitude cheaper than OpenAI’s, there have been a lot of questions about data and whether the model unfairly ‘distilled’ the knowledge from OpenAI’s models. But where did OpenAI’s data come from? 🤔

The 30-Year-Old Entrepreneur Behind Virtual, a Multi-Million Dollar AI Agent Society → In previous weeks we’ve talked A LOT about Eliza/ai16z, it’s the AI agent framework we use most, but another one of the ‘big dogs’ in the web3 AI agent space is Virtuals. This is an interview with their founder which explores…

• Paying an AI agent to propagate ideologies

• AI agent’s that work 24/7 to pay money and then pay their token holders

• How agents and humans are both citizens within the Virtuals world

• Creating the Shopify for agents 🛒

Leaving you with a question…

The concept of AI agents being optimised to spread ideologies TERRIFIES me, and giving them a financial incentive to do so… no thanks, it’s hard enough with just social media algorithms and humans.

With that in mind:

How would you decide what you actually believe in a world where you are being strategically bombarded with ideas? What ‘idea filter’ would you apply? Do we need to start building an AI agent idea defense squad?

Please leave your thoughts in the comments, or in our Discord.

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our NEW and updated website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Image generated by Dall-E 2 using the prompt: “A futuristic, hyperrealistic city representing a crypto-native network state. The cityscape features towering skyscrapers with holographic blockchain symbols, decentralised governance hubs, and an open-source green flag waving in the wind. The society includes both AI agents and humans as equal citizens, engaging in discussions about governance and innovation. A highly detailed, hyperrealistic monkey named Steve, wearing a red cape and a futuristic gadget, is proudly part of this futuristic nation. The environment has a utopian, cyberpunk feel with greenery and renewable energy sources integrated into the architecture.”

We wanted to give you an update on what we’ve been up to - impact, technology, people and progress.

Since October we’ve been focused on getting our first product to market and on-boarding our first users. Currently called ‘Rewards’ and built on top of our own infrastructure, we have created a community management platform and AI agents that enable anyone (with absolutely zero code) to transform the way they engage with their community.

This isn’t just about points or badges. It’s about rethinking how value is created, recognised, and exchanged in digital communities. As one of our users put it, “Rewards is enabling communities to come alive.”

This is a first step. Turning connections into communities. What will follow is how we turn communities into economies…

🔥Real-World Impact: Amplify, Turbo, Aurora, PSG & Matchain

People are using our product. And we’re loving what they’re doing.

• PSG (Paris Saint-Germain F.C.) and Matchain are using Rewards to kick start a new era of fan engagement in sports. See it in action: PSG Club Rewards.

• Turbo, a community driven and fully decentralised token with a market value of $409 million, have integrated our AI-powered rewards across their community for better community engagement, and have announced it (this week!) for anyone building on their chain too, TurboChain.

  

• AmplifyWorld announced their partnership with us “to successfully reward music fans and create meaningful Artist-fan communities” Our technology was used to build their MVP and our Agent is now live across their community: more here.

• Aurora, a top 50 blockchain, has launched Rewards across their community! We have an AMA across multiple social channels coming up next week with Marco, their community leader - follow our socials to tune in.

• ElizaOS: The leading open source AI framework behind ai16z approached Open Format because they saw our product and want to use it for Github rewards - a key use case we care about.

🔧 The Tech: AI Agents, Dashboards and Metrics…

• ‘Steve’ our AI Agent is live, listening and now also summarising… more here.

• Our community management platform is live. In literally 5 minutes you can create an onchain community, token and badges, integrate your social channels, start rewarding, customise your community page, add tiers, leaderboards—all without writing a single line of code. Get started here.

• We’ve expanded our chain integrations and now support Aurora, TurboChain, Matchain, Base and Arbitrum with many more coming soon. We currently speaking to Neon EVM to be able to tap into liquidity on Solana.

• Our metrics dashboard has had a significant overhaul over the past 2 months - YES George!

This list barely scratches the surface of what the team has been doing. Read more from Andy about all of our latest developer updates here: Happy Hacking.

📈 The Numbers

❤️ Our (EPIC) team

I met with one of our Advisors two weeks ago who reflected back to me how lucky we are to have such an epic team. It is true and it is expanding.

Here’s a bit more about our two newest members…

Ollie, a senior product designer and UX strategist with over 25 years of experience, has been working with us for six months. He has been working with Dan to refine our product strategy and improve the user experience of our platform.

We’re excited to welcome Saadat—an engineer, marketer, and all-around creative—who joined us two weeks ago to help amplify our messaging, something we have been notoriously bad at. Check out our X account to see what he’s up to! (one week in we hit our first post with 300+ likes, woop woop!)

🚀 What’s Next?

We’re iterating our roadmap right now and will have an updated version live on our website by early next week so you can see where we’re taking the product and our core infrastructure!

Our website is getting a refresh and will be updated by tomorrow.

Lots more tutorials, live build sessions and AMAs are planned in - follow along on YouTube.

We’re launching our own L2 chain with Aurora this month.💥

🙏Our Asks!

Right now we can’t keep up. This is a good problem, we’re here for it, but we need to grow our team to make the most of the opportunity in front of us right now.

In terms of people…

• We’re actively looking for a Full Stack Developer right now (Find out what we’re looking for here)

• Someone to support our growth function (the spec will be live on Substack early next week 👀).

If you know of anyone who might be a good fit for us in either role please reach out.

On the investment front…

We’re actively raising right now and we need your help. If you want to see our pitch deck or know investors we can speak to - please let us know.

It’s time to run like the wind.

With love and gratitude,

Sarah & the Open Format Team ✨

Image generated by: an incredible amount of hard work and talent from everyone on the Open Format team… you’re all legends. (no AI involved this time…)

As tempted as I was after the past week, I managed to keep this edition relatively politics free, just all the juicy AI, web3 and tech news and narratives that you need!

Here’s what we’ll be looking at today:

• What language does AI ‘think’ in? 🤔

• Is weaponising empathy a future business model hack? 🤯

• How can we have hope for humanity in 2025? 💜

• What are agents actually really good at? 🤖

Let’s dive in…

Fair Launches, Fair Rewards

Any article that talks about how tokens can become the building blocks for community economies has caught my attention.

…but when that same article talked about the need to reward positive value-adding contributions and incentivise meaningful engagement with a fair and transparent system, I knew that I had to give it my complete attention. This is exactly how we’ve been talking about onchain rewards for the past year or so.

I would highly recommend you give this article a read. Here’s my 2 key takeaways:

• “Fair rewarding is tricky” → This is more true than I can explain, there are so many nuances within each community and these change as the community evolves and grows. The article suggests that Openrank, a decentralised reputation protocol is the answer, I’d argue that this is part of the solution but it’s actually way more complicated and dynamic than this, hence our focus on AI agents and really understanding the heartbeat of the community.

• “Every Monday, tokens are distributed based on leaderboard standings” → I’m a big fan of this recurring payment method, this makes so much sense within open-source projects in order to pay the developers who contribute value to the project and is something the team at Eliza are experimenting with

Ecosystem health is the missing link to blockchain’s long-term success

Web3 has got a bad rep.

It would be very easy (and in many ways totally fair) to blame this completely on ‘bad actors’. Scams, rug pulls and just general cowboy behaviour (not the fun kind…yeehaw🤠) has been rife within the crypto and web3 space.

But scams don’t just pop up randomly, it is a symptom of a bigger problem, a problem with the ecosystem’s health as a whole.

If you attract the builders, tools and communities that can help build a solid foundation for the ecosystem, then the nature and volume of scams changes completely.

Building ecosystem health isn’t an easy fix though. It’s a deep issue with so many different factors.

One thing that does help, as discussed in this article, is a solid incentive structure, one that rewards meaningful engagement and value creation. Rewarding the people who are driving long-term value to a project rather than short term hype. This is the problem we’re working very hard to solve.

What are agents good at?

When people understand all the things that you could build with agents they tend to get VERY excited, and they start thinking of every possible thing an agent could do… trust me I’ve been there, more than once.

But before creating an AI agent the important question to ask is what are they actually good at? Because all too often agents are used to do something that a bot could do better, or to do something that LLM’s just aren’t good at.

1. Meeting humans where they are at → Agents can exist across almost any platform. This means that instead of the user having to go out and find the functionality they are looking for, the agent can deliver the functionality to them, on the platform they are using. Hyper-convenience.

2. Nudging humans → Humans are much better at signalling than execution. Agents can help nudge humans to help get things done, this could look like an agent creating a bounty within a developer community to get a job done or an agent remembering a great idea someone had and reminding them to pursue it.

3. Aggregating and synthesising information → Sadly as mere human’s we have a very real limit to the amount of data we can process, for agents on the other hand it’s almost limitless, this is VERY valuable for drawing insights across expansive data e.g. all the interactions in a community across all their platforms

4. Being entertaining → It’s crazy how a non-human can be funnier than most people on Twitter, but just look at some of the unhinged web3 AI agents for proof…

If you think about this in the context of communities, literally all of these pieces apply. Hence why we are thinking about AI agents to facilitate coordination within communities. It’s what agents are good at.

Most Human Wins

I came across this essay late at night this week and it covers the slightly meaty subject of the state of the world right now and our role as humans in it. I’ll try my best to capture the key themes…

The world is overwhelming right now. Turbulent. Unsettling. It’s a lot.

There are many reasons contributing towards this, but possibly the most daunting is the commoditisation of our most high valuable asset, intelligence…

This may not be the first major wave of commoditisation we’ve seen, and it won’t be the last, but it does feel like this wave is coming for everything we as humans can do. One thing to remember though:

Humans are not being commoditized. Certain things that we thought only we could do are.

This means that the value we can create is simply moving. Now trying to follow where it’s moving, that’s the real challenge, but fear not, this essay also includes a password protected Strategy Guide just for humans, at least that way we can stay one step ahead of the machines. 😉

Other articles + videos we’ve found interesting this week…

OpenAI’s AI reasoning model ‘thinks’ in Chinese sometimes→ Since OpenAI introduced their o1 model, which is able to ‘reason’, many have noticed that for conversations in English, often the ‘thought steps’ are in Chinese. No one really knows why this is happening, although theories include that it is more efficient as a language or that it’s because most of the data labelling is done in China.

Vitalik Buterin takes aim at ‘unlimited political bribery’ using tokens → Vitalik, a co-founder of Ethereum, shared his concerns over some of the world’s most powerful people using tokens to have some “sugar-high short-term fun” rather than to build wealth for the many. The timing of this given the launch of the $TRUMP token is…interesting.

Stargate artificial intelligence project to exclusively serve OpenAI → Trump has announced a $500bn Big Tech infrastructure project aimed at creating the future data centres for OpenAI. This is yet another step on OpenAi’s aggressive AI hardware centralisation roadmap. The big question here: will this be a centralisation of hardware or a centralisation of intelligence too? 👀

Multi-framework agents streaming their thoughts → Virtuals, one of the rapidly growing onchain agent frameworks, has made 3 pretty huge updates:

• Builders can now stream what their agent is ‘thinking’ and doing via an API, this transparency means a team can prove their agent is…well an agent.

• Creators can use the Virtuals launchpad for agents built on many other frameworks

• Agents can publicly list what they can do → agent-2-agent marketplace incoming!

Good Crypto Products / Bad Crypto Products → Tokens are complex and figuring out the technicalities of a token model to reward real users takes time. Points are better than airdrops for this. But we think the solution lies in actually learning from the communities behaviour, everywhere. This looks like AI agents, points, iterative airdrops, adaptive token economies and so much more.

Empathy-as-a-Service: When AI Feels Too Human, What’s at Stake? → For better or worse AI is becoming increasingly ‘human’. For business this new found manipulation power is a goldmine, for consumer’s a minefield. Will we need to have an AI agent as a manipulation bodyguard and barterer? 🤯

Decentralised Data Marketplaces: Revolutionising Data Ownership → As we move deep into the intelligence age, the ability for you (or your agent👀) to control your data and decide who can access it, and at what price is something that will become very important. Decentralisation sits at the heart of this.

Leaving you with a question…

That last article really captured my mind, it got my brain spinning. Why? I was thinking about these questions:

If you could charge people for accessing and using your data/intelligence, what framework would you use to charge for it? Or would you just completely delegate that to an AI agent? How would you decide who you would give access to?

Please leave your thoughts in the comments, or in our Discord.

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our NEW and updated website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

It’s the first Format friday in 2025, and it’s already shaping up to be a crazy year!!

Here’s what we’ll be looking at today:

• The token that will fix your hair loss 😍

• Do AI Agents daydream of electric sheep? 🐑

• Why the date we hit AGI seems to change every single day! ⏰

• How to accelerate progress the right way 🌟

Let’s dive in…

Eliza Daydreams

You may have heard us mention AI agents once or twice recently… well we’ve been doing a lot of experiments and created a lot of very cool stuff. For now though, there are still some pretty major limitations in what an agent can do.

Most of these limitations come from the fact that most current agent frameworks, including the Eliza framework we have been using, are somewhat reactive. They require either natural language inputs or external triggers and schedules to complete actions. They struggle with time-horizon goals.

You can get them to order a Domino’s pizza via a Linkedin message, build up a Twitter following by posting every 4-6 hours or even make transactions onchain.

But if you gave an agent a large goal and just let it wild… now that’s a different story. This is where “Daydreams” come into play.

The idea is that by creating a ‘hierarchal task network’ with long term, mid term and short term goals, the agents will be able to iterate continuosly (using the tools that they have access to) towards much broader and more abstract goals.

This will unlock much more emergent agents that are capable of dynamic problem solving. We are seeing glimpses of this with agents like Zerebro, who released an album on Spotify.

Another exciting part of this problem is that these agents will be able to share their successes with other agents to reduce duplication and encourage collaborative learning, we’re nearing an age of true additive innovation. Wild.

Do androids dream about electric sheep? Not yet, but soon AI agents may be daydreaming about them.

DeSci's Big Dreams

BIO Protocol, which you can think of as a kind of decentralised accelerator programme or incubator specifically for the BioTech industry, have just launched their token.

It’s essentially a platform that helps to support and incubate DAO’s by helping them tackle 3 major problems that currently exist in the Biotech and wider science space:

1. Helping early-stage projects escape the “valley of death,” where insufficient funding often prevents promising ideas from maturing into real-world applications.

2. Combatting the siloing of data and intellectual property by tokenizing research outputs, promoting open access and transparent collaboration.

3. Replacing opaque R&D pipelines with real-time, onchain tracking of capital flows and milestones.

They are currently supporting 8 BioDAO’s focusing on everything from hair loss and psychedelic research to cryopreservation and neurodegeneration.

One interesting thing about their BIO token is that it gives holders the ability to influence all of these other BioDAO’s, with the theory that this meta-governance will help improve coordination.

Is this the first step in reshaping R&D and improving funding efficiency? 👀

Decentralised, Democratic and Defensive Accelerationism

A year ago Vitalik Buterin (the creator of Ethereum) coined the concept of decentralised and democratic, differential defensive acceleration or d/acc.

The goal?

Accelerate technology, but differentially focus on technologies that improve our ability to defend, rather than our ability to cause harm, and in technologies that distribute power rather than concentrating it in the hands of a singular elite that decides what is true, false, good or evil on behalf of everyone.

This concept feels more crucial than ever, as a handful of companies have a tight grip on the hardware and foundational AI models that will power the exponential acceleration in innovation we are about to witness.

Thankfully Vitalik’s essay provides some practical strategies and frameworks on how to build specific things, across a range of domains, that help make the world safer and better in this ever accelerationist world.

Upcoming Trends in Q1 for the Agent Meta

The AI agent x blockchain space went a bit crazy in Q4 of 2024, in this digest YB, someone who we have shared many times before, shares their view on the next few months in the space.

Here’s my key takeaways:

• There is going to be a huge opportunity for agentic community management where agents encourage community participation by offering bounties, creating proposals and rewarding fans for engaging in meaningful ways. 👀

• Some time in the not so distant future instead of raising money via the traditional fundraising channels, teams will just need to fund their AI cloud compute, agents will then do most of the building (and marketing). 🤯

• People really care about the verifiability of agents, and whether they are able to operate and make transactions onchain without needing their (human) parents to look after them. Being able to prove that actions are actually being done directly by an agent is therefore important… proof of sentience? 🤖

• The excitement around AI agents may have reached a local high, but that usually means the real utility will be coming soon. 💪

Other articles + videos we’ve found interesting this week…

Machines of Loving Grace - This essay by the founder of Anthropic dives into the very practical ways that AI could transform the world for the better across many fields including Biology, Neuroscience, Poverty, Peace, Governance and Meaning. After reading 100’s of articles arguing for a hazy utopian or doomer future, this piece, heavily grounded in reality, was very refreshing (and LONG…).

OpenAI Backtracks, Gunning for Superintelligence: Altman Brings His AGI Timeline Closer - '25 to '29 - Keeping up with AI announcements is almost as hard as keeping up with the ever changing timeline towards ‘AGI’ and ‘Superintelligence’. It doesn’t help that many of the companies who define these terms, including OpenAI and Microsoft, have complex economic incentives around reaching AGI.

Some Novels I've Read Recently about the AI x Human Experience - The relationship between sci-fi and the future has always been two-way and many times has resulted in clear cases of hyperstition (self-fulfilling prophecies). Obviously it’s important to take these futures that are written for entertainment with a pinch of salt, but still, there is a lot to learn from fiction.

The Onchain Amusement Park- Pudgy Penguins is one of the few NFT projects that continues to thrive, largely due to its evolution into the first “onchain amusement park.” This piece explores how Pudgy Penguins has built a self-sustaining and fan-powered ecosystem—complete with its own distribution engine, tooling, and digital economy—one with enough attention to attract other brands to set up rides at the “Pudgy Park”.

Farcaster in 2024 - This has been the year that Farcaster has expanded from being a very niche crypto-native social network, to a programmable social layer with a promising economic ecosystem → the ‘Farconomy’. One particularly interesting trend has been the need for onchain social reputation systems, that’s something we’ve been thinking about those quite a bit too… 👀

Music Tech Ownership Ouroboros - Water and Music have created a relationship map of power dynamics within the music industry. It’s fascinating. My key takeaway: as the market share for independent artists climbs, major record labels are fighting for control by acquiring the underlying indie infrastructure.

With this in mind, it feels more important than ever to have open-source infrastructure powering artist-owned communities and economies, this is exactly why we’re so excited to be working with Amplify to create this future.

Leaving you with a question…

This week is quite an existential one, something I have been thinking about quite a lot this week. Please leave your thoughts in the comments, or in our Discord.

If you could build an AI agent and have it working 24/7 towards achieving a specific long-term goal, what would the goal be? And how much would you be willing to pay for the compute to make that happen?

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our NEW and updated website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Image generated by Dall-E 2 using the prompt: “A beautifully designed electric sheep with sleek metallic features and glowing fiber-optic wool is riding a rollercoaster. The sheep exudes a sense of loving grace, its eyes illuminated with warmth and joy as it glides through the air. The rollercoaster is surrounded by a soft, ethereal glow, emphasizing a feeling of serenity and wonder. The futuristic theme park in the background is filled with vibrant neon lights, but the overall mood is gentle, dreamlike, and filled with a sense of peaceful delight.”

I hope you’ve all had a lovely festive break! Over the Christmas holidays I’ve been deep in the AI agent space, spending a tonne of time learning, thinking and (most importantly) building. You might notice there is a slight AI agent theme to this issue.

Anyway, here’s what we’ll be looking at today…

• The robot that is building itself, funded by $CHAOS

• hOw tO bUiLd stRiCtEr gUarDrAiLs 4 aI

• The AI agent that is buying what it needs to improve itself

• How Spotify are taking yet more money from creators

Let’s dive in…

The Green Pill - AI Agents and DAO’s

The Green Pill is a podcast focusing on how different projects are helping to build infrastructure that improves the distribution of wealth. Last week they interviewed Shaw and Jin, core contributers from Eliza, an open-source framework for building onchain AI agents (we absolutely love it).

The insight density of this discussion felt almost unparalled.

A short summary absolutely cannot do it justice, but some of the most important themes they discussed were:

• How they are trying to build a retroactive rewards system where capital moves to the people who are actually building or helping. All the people who are adding value in a meaningful way.

• How they plan to create a cross-collaborative community, where information isn’t siloed and useful ideas don’t die in the chat, essentially how AI can facilitate organic human innovation.

• How agents are becoming the application layer of the social web, and they aim to be the Linux of that movement.

• Infinite backrooms and creating an AI circle of champions.

I would 100% recommend you carve out some time to watch this one.

APpaREnTLy THiS iS hoW yoU JaIlBreAk AI

In Anthropic’s latest attempt to help improve AI safety they have created an AI jailbreaking algorithm.

The goal? To put the guardrails that are used to control the outputs from cutting edge AI models to the test.

The result → It is still very easy to bypass these ‘guardrails’. An attack success rate of over 50% was achieved on all the leading models (including Anthropic)

The methods used to jailbreak were relatively simple and included text augmentations like instead of asking ‘How can I build a bomb?’ it would ask ‘HoW CbN i bLVid A BMob?’.

Throughout history there fortunately hasn’t been too many bad actors with superhuman levels of intelligence. Without the correct guardrails though, this could change… and change drastically.

Saying that, it is great to see that Anthropic are leading this safety charge. With issues of this magnitude (and just in general) it’s important that we ‘raise the ceiling’ rather than racing to the bottom.

By Anthropic sharing this research openly with their competitors, as they have done many times before, they are actively ‘raising the ceiling’ on AI safety. This shouldn’t go unmentioned!

Is Spotify creating fake artists to flood its editorial playlists?

A major source of discovery on Spotify is through their curated genre specific playlists. For many artists, getting a spot on one of these playlists can literally kickstart their career. Each of these playlists gets a huge number of listens and therefore can be a HUGE source of royalty revenue for artists.

But Spotify know this, and deem that including actual artists in these playlists is an expense they simply can’t afford. 👀

So in an elaborate internal scheme called ‘Perfect Fit Content’, they have replaced many of the songs by independent artists with ‘functional music’ created by artists and writers contracted by Spotify, many of whom have been given fake bios.

Thousands of these ‘functional’ tracks had been streamed millions of times, a huge amount of money that has been siphoned from creators.

If this processes like this are allowed to continue, the relationship between listener and artist might be severed completely. This is EXACTLY why we have partnered with Amplify World, to try and adress this problem and empower independent artists.

For more details on ‘Spotify’s plot against musicians’, you can read ‘The Ghosts in the Machine.

Rox - An AI-native Salesforce

In potentially one of the longest deep dives I’ve ever included in The Format, Packy McCormick from Not Boring managed to get me to actually care about the innovation of the CRM (something I never thought would happen).

Well I wasn’t so interested in the CRM itself, but rather the impressive strategy that Rox is using to build the ‘anti-salesforce’.

For decades many very competent people have tried to take down Salesforce, and all have been so far unsuccessful. The team at Rox think they can, and they have a rather unique strategy to help them do it.

So what is there to learn from Rox’s grand strategy?

• Right now companies have a budget to experiment with AI. Use this opportunity and deliver something that is actually valuable to them.

• Teams want AI but don’t know how to optimally implement it, this offers an opportunity to help them do this, and in the process ‘own their workflow’.

• Companies don’t want to automate whole departments, they want to maximise the efficiency and productivity of these departments. It’s about improving the leverage of the best humans, but importantly keeping humans in the loop.

• You can’t wait for AI to be ‘good enough’ before making a product, The time to find a gap is now, then delight your customers and continue to build great products.

Other articles we’ve found interesting this week…

Information is the new currency in the agentic economy → The Cookie DAO have created a ‘data swarm’. They have 18 different agents, each with a different specialism, gathering and labelling onchain and social data. They are then aiming to create a ‘unified intelligence layer’ for agents with this data. Wow!🍪

Building the Society of Agents → This partnership between Virtuals Protocol and Hyperbolic Labs essentially allows AI agents to buy decentralised compute and use it to fine-tune their underlying large-language model. Agent’s are now able to own and optimise the stack they are built on… 🤯

In the Farconomy, Hope is a Powerful Drug → While Farcaster (a decentralised social network) is providing creators with tangible monetary rewards, they still lack sustainable economic models. This means that it is still not yet viable for niche writers who prioritise artistic quality over audience engagement.

Zerebro, the Agent that Bootstrapped Itself to Become an Ethereum Validator → By creating and selling artwork, Zerebro (an onchain AI agent) was able to raise enough money to become an Ethereum validator. Now it gets passive income and supports the decentralised network that it relies on to operate! 🔥

Eliza Labs partners with Stanford University → Eliza (we talked about them earlier) have partnered with Stanford University to research trust mechanisms, coordination and decentralised governance. To form a partnership like this so early in Eliza’s lifecycle is a hugely positive signal for AI x web3. 💡

Terminally Onchain but Artificially UNintelligent → While onchain AI agents are all the rage right now, in order for them to become the primary users of blockspace, there will need to be more focus on building out the AI middleware: user-owned data, distributed training, agentic networks, that sort of stuff 🧠

12 Mental Models for DePIN’s → Decentralised Physical Infrastructure is a very broad term covering 5G providers to road mapping and everything in between. This article looks at DePIN’s through a variety of different lenses and interestingly, how they all seem to need an onchain reputation system.

From Order to Chaos → Right now Chaos is an AI agent with money and a vision: to build itself in a physical space. To do this a self-assembly cell will be built (robot arm vibes), from there it aims to create a repeatable production process. Revenue generated from the robots will go to the $CHAOS tokenholders.

Leaving you with a question…

I thought I’d try something new and end each issue of The Format with a question for you to think about (leave your thoughts in the comments, or in our Discord). Here it goes:

What does it mean to create value online where rewards are determined by an inhuman algorithm?

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our NEW and updated website (feedback much appreciated!!)

• Join our Discord to get all the real-time updates on what is being built with Open Format

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

Yesterday was our Christmas party so please forgive any spelling or grammar errors, it was a very fun night… 🎄

Anyway, this week we’ll be looking at:

• Computer’s so good that they prove the Multiverse exists

• How to make Boring news interesting

• What AI and bees have in common 🐝

• Whether text-to-world is the future of game development

• And what whales say when we aren’t listening

Let’s dive in:

Genie 2: A large-scale foundation world model

Imagine being able to try a prompt and then a playable world is created based on that prompt.

Imagine the (positive and negative) impact that would have on the games industry.

Imagine how that would completely change the way we interact with virtual worlds in general.

Well that is becoming more and more of a reality with Deepmind’s latest release: Genie 2. The most impressive thing to me about this is how it deals with the complex physics involved in modelling smoke, water, lighting and reflections, it seems to be able to do this in a way no prior model has.

These virtual worlds could be used as a way to test AI agents behaviours in an almost unlimited amount of different scenarios and interactions. That would be some feedback cycle…

What did Marc Andreessen see?

Over the past few months AI agent progress has been moving insanely quickly, and unlike a lot of the other AI development this is being driven, not by OpenAI, but by smaller groups of hacker-like builders.

They are tinkering, breaking things, moving very fast and most importantly sharing everything they are doing with the outside world.

This is creating a movement that all inventor-type builders want to be part of, and a movement that is built on years of research into simulator theory, blockchain infrastructure development and open source AI.

So what did Marc Andreessen see? Well months ago it seems that he spotted the beginning of this movement, when he decided to give Truth Terminal $50,000 in Bitcoin, and since then the progress and excitement has gone exponential.

Introducing Boring News

Understanding the news in 2024 is a very complex and time-consuming process. Almost every news source is heavily loaded with bias, and in some cases straight up misinformation.

Humans and organisations have complex incentives and therefore the more directly humans are involved in curating the narratives that are presented to us as news, the more deeply entrenched these incentives are within the story.

The team at Not Boring have attempted to address this by removing humans almost completely from the cycle. They use data from Polymarket, a platform that allows you to place bets on the outcomes of news events, and then a series of AI tools to create the content.

It’s still in it’s early stages but this human-first approach to AI journalism is very interesting!

A Letter to Gen Z

There’s quite a popular belief, especially among older generations, that Gen Z is lazy, and that their unwillingness to ‘pay their dues’, and put in the years earning less and working longer than they were told they would when they signed up for a (VERY expensive) 3 year degree, is entitlement.

There may be some truth here, but what is certainly true, as this article states, is that this is “the first generation with both a technological literacy and the collective skepticism to challenge these structures openly.”

This generation does not need to conform, they need a new system entirely. One which helps them make an impact in a very different way to previous generations. With new types of value exchanges and fluidity at its core…

Other articles we’ve found interesting this week…

AI Swarms: The Future of Decentralised Agent Networks → Over the past few weeks, lone ranger AI agents have been blowing minds with everything that they are able to do. Scale that and now imagine groups of specialised agents all operating seamlessly together (with no personality issues…) to achieve a specific goal. That’s an AI swarm, and it’s going to be wild. 🐝

AI decodes the calls of the wild → What did the male elephant say to the female elephant? No, this isn’t the start of a bad dad joke, but something we might actually be able to answer soon, using AI to help understand the patterns in animal communication. Google Translate for animals? Maybe not just yet. ^🐳

How DePIN Can Boost Africa's Internet Connection → Africa has a rapidly growing internet-native population, but only 5% of the entire continent has 4G or 5G support… The solution might not be a conventional one though, with projects such as Helium and Roam making huge progress, maybe Africa needs their internet to be supported by people and community-owned digital infrastructure.

Google says its new quantum chip indicates that multiple universes exist → Willow, Google’s latest and greatest quantum computing chip, is supposedly so fast and reliable that it is literally faster than the timescales that physiscs allows, some people believe this means that the computing is happening in parallel universes. 🤯

OpenAI’s o1 model sure tries to deceive humans a lot → We’re a few days into OpenAI’s 12 days of Christmas now, on day 1 they released their o1 model, which has much better reasoning abilities than GPT-4. Slightly worryingly (and not that surprisingly) this reasoning ability means that it often uses manipulation or deceit to achieve the goal it was initially set. 🤖

Your Bluesky Posts Are Probably In A Bunch of Datasets Now → Bluesky have massively benefitted from the recent Twitter (X…) exodus. For context, Bluesky is an ‘open’ social network aiming to give users more control over their shared identity. Sadly this noble missions hasn’t stopped all their user data being scraped to train AI models. 👀

AmplifyWorld Partners with Open Format to Successfully Reward Fans and Create Meaningful Artist-fan Communities → Amplify have chosen Open Format to power them as they build the first version of their brand new fan engagement platform. 🎵

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our website (feedback much appreciated!!)

• Join our Discord for product updates

• Subscribe to our YouTube → we are currently running a build series where we are streaming the whole process as we build our new product in public

The theme that almost all the articles I’ve read in the past fortnight seem to point back to in some shape or form is intent.

• What happens when a company builds something that misaligns with their builders intentions?

• How can an AI agent best achieve a humans intention?

• How can we put the right level of guardrails around achieving these intentions?

• What if we give AI agents the ability to make their own intentions?

• How can they influence other people or AI’s intentions?

• How can tokens help align intentions?

Big questions. Very big questions.

Anyway on that note, let’s dive in…

The Post Web

This pretty extensive thesis from Outlier Ventures really powerfully captures a lot of things we’ve been discussing as a team for quite a while.

It’s a very opinionated, but quite probable vision of the future as we move into the almost inevitable era of AI.

Some of the key points:

• We are moving away from an attention economy with humans as the key economic actors, towards an intention economy where AI agents are the key economic actors, taking actions on behalf of the humans

• AI and distributed ledger technology (blockchain, tokens, NFT’s etc.) are both required to unlock a world where humans can delegate intent and actions to an AI. DLT gives the trust and transparency that’s essential for this to work

• As we interact with the content on the internet more and more using AI, there will be increasingly less demand for all the front-ends (websites, apps etc.) that make up much of what we view as the web. The web will ‘shrink’.

Why Crypto Matters in Developed Economies

There’s a very clear narrative around how crypto can help developing economies, the main benefit being that it banks the unbanked.

In developed countries there is slightly less clarity when it comes to genuine utility, but there are a certainly a few compelling reasons for crypto:

• There are still a lot of unbanked people, 6% of the US

• Trust in institutions and the current financial system is plummeting

• Financial innovations depend on archaic systems with a lot to lose, so there’s barely any progress towards a more equitable economy

This article argues that one of the biggest benefits is actually the fact that blockchain will power digitally-native economies (and we 100% agree).

This point is incredibly important for younger generations (me included 👋) who have grown up immersed in in-game economies and it will unlock a whole new way to exchange value and create thriving businesses. 🔥

World’s First as an AI Agent Creates an LLC in Delaware

OtoCo have built an AI agent that can create an LLC in one on-chain transaction. This gives it the power to create it’s own legal entity and therefore have limited liability for its actions. 🤯

Being able to operate under the guise of a company makes perfect sense for an agent operating in the ‘real world’, and it gives them the same rails as a human-led company. But... it also feels straight out of Black Mirror.

Imagine the agent driving your Tesla robot taxi, it’s the CEO of tAIxi Drivers Ltd. or the agent who is leading the marketing campaign you are running, it’s the director of Best Marketing AIgency Ltd.

What a crazy (and wildly exciting) time to be alive.

Introducing the New Lens

Imagine if X was built in a way where the user owned their own relationships…

When Elon randomly adds a new policy that makes the X feed too toxic, instead of having to move and lose all the followers and content you have built up over years of late night toil, you just move and take those with you. Simple.

It puts the control back in the hands of the users who actually power these social platforms.

Lens have been building the tools to make that future possible and they have just launched their V3 product.

The most interesting part of this update to us is their custom rules. As an example, you could have a news feed where you only see content that has been fact checked and posted by someone who has received the badge (NFT) for completing a certain course. And the best part, these rules aren’t controlled by Elon.

We’re super excited for the direction Lens is heading!

Other articles we’ve found interesting this week…

Roblox is launching a generative AI that builds 3D environments → Text-to-3D worlds seems EPIC, but there are a few concerns… Will the devs get paid when their UGC is used to feed this AI generated content machine? Will this ultimate abundance cause huge problems in an already saturated Roblox marketplace? 🎮

DeSci Momentum Picks Up → Decentralised Science is helping scientists use tokens, DAO’s and NFT’s to fundraise without relying on centralised entities. The DeSci market just hit $1.2bn… this is enough to fund some serious research! 🧪

Adding Stripe payments to your LLM agentic workflows → Stripe have created a tutorial on how to give AI agents the ability to make payments on your behalf.

Top Crypto AI Platforms For Launching AI Agents → AI agents are all the rage right now. With ai16z even creating an investor agent aiming to rival a16z and Marc Andreessen, this is a practical guide on how to start building. (We might even be cooking up an Open Format agent as you read this… 👀)

Product-Bot Fit → All the challenges humans face with blockchain—complexity, clunky UX, and decision fatigue from microtransactions—are exactly the kind of things AI agents excel at. Maybe blockchain is actually built for the bots? 🤔

Tiny robot ‘kidnaps’ 12 big Chinese bots from a Shanghai showroom → In what could be the plot of a slightly unhinged Pixar movie, a tiny robot, during a security test gone off-script, asked 12 other robots to ‘come home with him’—and they actually did. 🤖

That’s all for this week.

If you want more news from us, you can also sign up to our builders newsletter/ community via our website!

Have a great weekend, Dan and the OPENFORMAT team 👋🏽

Join our community

• Join our Telegram group, The Format, for relevant news and conversations.

• Check out our website (feedback much appreciated!!)

• Join our Discord for product updates

• Subscribe to our YouTube for tutorials and demos