Tlon's Plan

Galen Wolfe-Pauly
illustration of grid of dots

Tlon’s initial product is a high quality messenger for small groups. Over time we’re going to evolve that messenger into a social OS.

Most of what we do with social apps and services can be much more effectively done inside a unified, high-quality system that’s entirely user controlled and freely extensible. We’re going to get there by first building simple tools for a small (and reasonably technical) niche. Then we’ll expand outward toward a multipurpose platform.

Our goal is to build the best platform for staying connected in the world by pioneering an entirely new category: software you can trust completely.

Why can you trust it completely? Because you own all of your own data, run your own apps, and communicate directly with those you want to connect with. Our business has nothing to do with spying on you and never will.

We think that a high-trust platform will have a positive impact for everyone. The more that digital communities can trust their tools, the more they can pioneer new ways of seeing the world, collaborating and building things.

Today, everything we do is owned and controlled by some company. They're mining our data, using it to train models, sell us more stuff we probably don’t need, and keep us engaged. Above all, it doesn’t feel good to live in a digital world controlled by someone else. We think being able to trust our tools to stay connected will make everyone more optimistic.

This post covers our strategy and it could very well take us a decade to complete. We're clear about the near term. As the plan gets further out it’s less high resolution, but we’re still pretty sure about the broad strokes.

Let's walk through why we're doing this, what the road ahead looks like, and the first steps to realizing our vision.


Everything about the current system for communicating and staying connected is structurally broken. The digital world is an essential part of our lives, and the whole thing—from infrastructure to interface—is controlled by a handful of companies.

There’s nothing wrong with companies, but centralization of power over digital communication does present significant risks. We can be manipulated, censored, quietly coerced and mined for attention. All of these things have already had a well documented impact on society. Their future impact could be even more dystopian than we can imagine.

The risks of the current system tend to sound abstract, but they’re far from it. The photos we share with friends and family all live on a server run by one of these few companies. It’s extremely unlikely they’ll last forever, perhaps not even as long as our photo albums from the past. Our teams collaborate using tools that can be compromised without our knowledge, leaking all of our information to the public. Or, similarly, we could lose our entire work history without warning. The manipulation of free speech, especially political speech, is well documented and against the goals of a free society.

Beyond any of these individual concerns, though, is a missed opportunity. Without the capacity to freely and conveniently explore the most powerful technology available to us, we are never able to realize the full potential of a general purpose computer with a network connection.

We want to have fun. We want to be able to experiment, build, and connect in a totally permissionless, unrestrained way. We want to be free. We want pure, powerful tools that we can do whatever we want with, across any distance, between any two points with an internet connection (which is pretty much anywhere). We want all of that in a convenient, accessible package that’s easy for anyone to use.

Building a better system for people to connect and communicate has the potential to impact everyone, everywhere. Our bet is that when communities control their own tools, they can coordinate and collaborate more effectively to pursue their own definition of progress. Let’s talk about what that looks like.

How? Start small, expand

No system is built all at once. There are three steps to delivering a social OS:

  1. Build a personal server
  2. Build an intelligent messenger for groups
  3. Evolve that messenger into an extensible OS

Step 1 is the prerequisite problem for everything we want to achieve. We spent the first decade of Tlon’s history building Urbit as the solution. Urbit is a complete, wholly encapsulated system for a user to run a personal server in the cloud. We built Urbit to deliver a well-crafted, maintainable system for individual cloud computing far into the future.

Rather than get too in-depth about Urbit as a piece of technology, the important thing for understanding what Tlon is up to is to understand how categorically different Urbit is:

In the current world we live in, everyone goes to their software. We connect to the software provider, who stores all of our data, runs the app, and controls everything about it.

In an Urbit world, our software comes to us. Each Urbit user runs their own node that stores their own data and freely networks with other Urbit nodes as needed. The network is maintained as an open system and protocol and owned by everyone who depends on it.

(If you’re curious to dive into Urbit as a piece of technology, head to the site.)

This is a new paradigm for individual computing in the cloud. Your Urbit node belongs to you, can run anywhere, and can conceivably last forever. The entire system is open and can be audited; the network is both encrypted and peer-to-peer. Urbit provides the guarantees of openness, ownership, and durability—what we expect, but don't have, from a tool for staying connected.

Urbit on its own is just technology. We want to materialize the power of that technology for everyday use. While Urbit isn’t finished it is good enough for step 2.

Step 2: A better communication tool

We're starting with a high-quality messenger for groups.

Messaging is a constant part of our everyday lives online. Most of us are messaging all day, every day. And yet, our options for messaging are really limited.

"There are already so many messengers," we often hear. Yeah, we know. We think they're all lacking, compromised or frustratingly rigid. They're all going to disappear, none of them are extensible, and only one of them is probably private.

Apple keeps us permanently locked in. Signal is great but narrow and will always rely on the Signal Foundation (and you still have to trust the Foundation). Discord and Slack are chaotic and stressful. When you customize them with bots you leak data to even more third parties. Telegram is fast and responsive but has a deeply suspicious security model and no hope of lasting for the long term.

A messenger is exactly the kind of tool that we should expect to be as simple, durable, and high-trust as your dining room table. A messenger should be warm, inviting, and get out of the way so you can build a sense of genuine connection.

That's what we're building.

Our messaging app is completely open-source, decentralized, and truly peer-to-peer. You can audit what it's doing, and it's on par with the tools we're familiar with. Fast, responsive, and a convenient alternative to the services currently mining our attention.

We’re going to go much further, though.

First, with our system you don't just get independence from existing platforms, you get a level of durability that isn't offered by anyone or anything that exists today. Our software will still work even if Tlon disappears.

The promise of being able to use software even when the company making it is dead means you can trust the tools we make beyond any other option. We can make this promise because what we've built runs on top of an open platform that can be freely run or extended by anyone.

Second, our system is one where all of the chatbots that augment your day-to-day work can effectively run on prem. They run on your own node, against your data, without ever calling out to the developer. This is a vastly superior model to anything that already exists—both for users and developers.

Developers never have to run a server, custody user data or think about configuring a stack. You can just build bots and ship them to people who run them independently. This means you can also build bots that have access to the entirety of a user’s conversation history without worrying about the security overhead.

For users, you can freely experiment with the bots, extensions, and plugins you want to run without wondering where you’re leaking data. For technical teams and communities, this is a much better paradigm.

This ‘everything on prem’ model is the seed of how we’ll evolve from a messenger to be a completely general system for collaborating, communicating and, in time, doing business since each user can freely run whatever software they like.

We’ll begin previewing a first version of this app to our alpha users this summer and open it up gradually. Toward the end of the year we’ll start showing off the first bots, extensions and plugins.

Step 3: A social OS

In time, our messaging app will expand into a social OS.

The PC was rightfully often talked about as the bicycle for the mind. We’ve spent a long time pushing that very single-player metaphor forward without reassessing it for the networked age. We have bicycles for the mind, what is the bicycle for the mind on the network? The social networks, apps and services of today surely aren’t the answer.

Messaging sounds like a boring, well-known problem—but if we’re going to build a more efficient system for people to compute socially, it’s the perfect starting point. From messaging we can expand outward toward a very powerful, network-native system.

Most of the apps that we use to connect with others could, instead of apps, simply be extensions or plugins. Slack and Discord bots demonstrate that this can be really powerful in a very narrow, chat-based way. The super apps of Asia push it even further. We can go beyond both of them.

We look forward to a future where adding a plugin to your team’s profile can turn an image channel into a store for your fans, an app can set up a booking system that you can customize, and a chatbot will stream your sales updates directly into your team’s chat. You’d buy these plugins directly from the developers building them and never leak data to them.

The blog you run with friends as a side project would publish through the same unified system. You could gate access to subscribers with an extension, run a chat for premium members with another extension, and keep up with your collaborators all in the same place.

Going further, our OS is the natural place to interact with LLMs. Instead of having to commit myself to the products of a giant company (and compromise everything about your privacy and ownership) to access their models, we can let you retain ownership of all of your data and software and run a model as a bolt-on augmentation. The future in which we're supported by fully autonomous AI agents is one to look forward to with the platform we're building. Less so if we are wholly owned by giant companies.

The system we’re building is one that dramatically lowers the cost of running custom software for a user and the cost of shipping software for a developer. Users can freely customize the software they run because they’re not locked in to their tools by identity, social graph, or data storage. Developers can much more efficiently ship software because they never have to run a server, build an interface, custody user data or configure a stack.

User and developer experience is categorically better on a network that’s fully decentralized, owned by its users and is entirely peer-to-peer.

Most importantly, though, no one is expected to care about any of those things per se. It’s just really nice to use and build for. Our social OS lets us freely explore what’s possible through being connected with general purpose computers. We’ll look back at the world we have today and wonder how we endured it for so long.

It’ll take us some time to get to this end state. Chatbots and plugins are a good starting place. But we have a very unique opportunity to shift the way the world is connected. It’s going to be great fun to go after it.

What’s next

We’ve experimented in the past with prototypes, experiments and betas.

2023 was Season One for Tlon. We learned a lot, brought a few thousand people into an early version and made huge technical progress.

2024 is Season Two. We started fresh on many fronts, rebuilt huge areas of the system (and team), and will have a version of Tlon ready for small groups this summer. We're people who like to build things, and we made the communication tool for people like us: who value what they make, who are optimistic about the future, and who want to explore every possibility.

We look forward to sharing our work with you.

In the meantime, if you want to join us on our mission, please reach out ([email protected]). Otherwise, sign up for our mailing list or follow us on X to keep up with our progress.