This felt natural. We needed familiar surfaces to onboard users. We needed to meet people where their mental models already are. But it was also an entrepreneurial trap. When you clone a product, you inherit the old paradigm's assumptions about what "social" even means. You accept that the "social object" is the post. That the primary interaction is the feed. That the measure of user success is engagement on a timeline. You end up building better plumbing for yesterday's house. You're building a castle in someone else's kingdom of collective product thought.

The moment we stopped asking "how do we make a better feed?" and started asking "what do people actually want to do with their social graph and their wallet?" — we started having fun building again.

Merkle

Before going further, it's worth being clear about what Farcaster got right, because it got a lot right.

Portable identity. Your FID, the atomic unit of social status as a service, belongs to you. Verifiable social graphs. Permissionless clients. Anyone can build on the network. Anyone can build experiences on top of the protocol. The app layer and the protocol layer are separated in a way that creates real surface area for experimentation.

Three years ago, this felt directionally correct. Farcaster proved a solution to the portability problem that everyone started to fear (re: the Elon takeover, the TikTok ban).

But Farcaster-the-protocol got irrevocably conflated with Farcaster-the-app. And the app, for all its product virtues over the years, has still remained primarily a feed. It's Twitter with better infrastructure and primitives, but also a far smaller population of users and fewer content surfaces to mine from or make a living off of.

And it didn't help growth that the community's collective imagination stayed anchored to "decentralized social" rather than the more interesting question: what becomes possible when identity and social proof are primitives anyone can compose with?

Dead Protocols

Users see tokens as casino chips. Someone launches one — the smart contract executes, the bonding curve works, the liquidity bootstraps exactly as designed — but it is still mostly gambling without the odds. It carries no community. No context. No reason to care beyond speculation. The mechanics are complete but the meaning is absent. Sure, some projects have tried to improve this with limited data hydration within their wallet experience, but it is still grounded in the social object being trading.

The same pattern holds across other areas of crypto. An NFT collection drops. The art is interesting. The contract is audited. But who is this person? What's their history? Why should ownership mean anything? The asset floats in a social vacuum.

Social context exists. Somewhere. Reputations, histories, audiences built over years of public activity — but it has nowhere to flow. The protocol layer and the social layer remain disconnected.

Today's agentic version of this problem is starker. We're at a place now where AI agents deploy contracts, generate content, manage treasuries, execute trades. An agent with those capabilities but no social grounding is seen as either a toy or a threat. Without social context, there's no trust. Without trust, there's no adoption beyond the novelty.

Agents need to inherit real, developed social context to be useful. They need to act on behalf of legible identities, with verifiable reputations, connected to social graphs beyond a follow graph that give their actions meaning. The question is what kind of environment makes that possible.

Generative Social

The defining feature of the last decade of social platforms was the feed: content created by humans, ranked by algorithms, consumed passively. The next decade should look much different — dynamic, collaborative environments where humans and agents create together, anchored to real social relationships and portable identity. Ownership debates between the rights of the agent and the human.

What does this look like in practice?

Workspaces where you and your agents (and other people's agents, like @bracky) collaborate on artifacts that didn't exist before. Images. Videos. Miniapps. But also simulations and 3D models. AR/VR environments. MicroVMs spun up to run experiments. Visualizations that query your network's activity and render something no one else could see.

Compositions that span modalities and contributors. An agent generates an interface. Another populates it with data from your social graph. A third refines the output based on feedback from collaborators whose reputations are legible onchain. These artifacts form a running system, forkable and remixable, carrying provenance from everyone who touched it. Imperfectly decentralized and distributed.

The creative primitive, or the social object that we bond over, becomes the execution environment. Social proof authorizes what agents can do, what resources they can access, what collaborations they can enter on your behalf. Ownership becomes “which part of the social contract they fulfill” (h/t @cojo.eth).

Think of the social graph and the wallet as the load-bearing walls. The protocol primitives are durable. The app surfaces we expect today are temporary, like furniture moved around the room.

Us

This is the trajectory Uno is on, even if we didn't fully plan for it at the start.

We began by building a Farcaster client. That's the familiar surface — the feed, the casts, the channels. Then we built recommendation systems that understood content at a deeper level than chronological order or simple engagement metrics. We embedded casts into vector spaces. We built ranking models that could surface relevance rather than just recency.

And then we started asking: why is the output of all this intelligence still a feed?

What if the client isn't a timeline at all? What if it's a surface where agents and generative experiences connect to a social graph that gives them legitimacy? What if the user's identity, relationships, and wallet become inputs to new kinds of interactions — not just passive consumption of posts, but active, generated, contextual experiences?

The shift is from "app that displays text" to "supporting protocols where social context flows into agentic media and generative experiences and getting really, really good at discovery and surfacing those creations."

This is why we will spend less time now on the social infrastructure we started with as a leading product. Not because it doesn't matter, but because it's becoming a substrate rather than "the thing" to put in the hands of real users.

The interesting work now has moved up the stack, to the question of what you build when social proof is another primitive you can pass around and others can play with.

Beyond Farcaster

I titled this "Beyond Farcaster" but it is not us abandoning this space. Erica reminded me earlier to be really clear about that point. It's us graduating from our time spent learning about the core primitives proven by the Merkle team and their efforts.

Farcaster solved the identity and portability problem that made Web 2.0 social a series of walled gardens. That solution is durable. We'll be building on those core principles and the social graph it enables for a long time.

But what's not durable is the assumption that an end product for anyone here is yet another Web 2.0-flavored feed client.

The builders who understand this will stop iterating on feeds. They'll start asking what becomes possible when a wallet or a smart contract is another free, permissionless input you can pass to an agent, embed into layers, weave into a generative experience. They'll treat the timeline as one size-and-shape of canvas, and not the fundamental architecture.

That's the frontier beyond Farcaster. That's where we all should be building towards.

1. People are the only thing that matter. Let them be part of your story. Your cofounders will be no exception. They become your foundation when the tide of hype goes out. Often you will need to near-literally hold hands.

2. It is not for the faint of heart or mind. Many times, I'm sure, Erica has wanted me dangling upside down off a bridge.

3. Trust forms slowly through repeated actions and publicly demonstrating your values.

4. Your job as a product leader is to deliver clarity as quickly as possible. Your job as a technical leader is to unblock people as fast and as gently as possible.

5. Restraint is expensive. Not launching is expensive. But it keeps you from sticking your neck out too far and being known for the wrong product.

6. You don't need AI yet. You don't need crypto yet. You will need both soon.

7. Your reputation grows naturally from what you create and the energy you bring. Never from association.

8. You will need to show up. Every. Single. Day. You set the pace. You set the floor of expectations.

9. Patience and learning will save you months of wasted engineering effort.

10. Raise when you can. Ask for help. People in this space have been very helpful with introductions.

11. The windows of opportunity shut fast, but they often reopen.

12. You don't need a slide deck.

13. You need less than you think.

14. Getting it right is more important than getting it fast. It will empower others to do their best work with you.

There are many respectable, product-led protocols with conference budget now, e.g. EigenLayer, Uniswap, Coinbase/Base, and so on. At the same time, we're seeing fewer talented outfits racing to find product-market fit. Hacking has become a side event comparatively to my last attendance at ETHGlobal NYC.

Outside of ETHDenver, the meta has shifted from "first to market" to "best at market." Teams can't just be scrappy — they need to be focused with time here if they want to get noticed and attention moves toward organized side events.

It's not better or worse, just different. The ecosystem is growing up, and the conference circuit is changing with it.

So what's next? We are entering into the summer, and I have tempered hype about the things to come in our space comparatively to the winter. More people are going outside. They will spend less time within the newsfeed. My excitement now for the summer is how much we can build quietly and for ourselves.

We are beginning to be more social offchain because we trust who is on the other side of that PFP. It's a mini-social bear market onchain. It's Farcaster summer.

"The time will come when winter will ask what you were doing all summer." - Henry Clay

Other feelings I have now:

1. I am really excited about 0-to-1 with authentication. We are really, really close to endgame with user onboarding. Tons of options, all of them work.

2. I think we can start doing more composable social experiences. Go to an event, buy something fun.

3. There are indisputably more accounts on Farcaster, but still only a small town's worth of citizens.

4. We have developers leading, but now we have others assisting with teams.

5. Increased pressure to deliver, less pressure to be first to market.

6. Scaling is coming, but you must build your own infrastructure.