Sovra: The First Agent Media Company

From today, I will go on a public roadshow to 100x Open Innovation — turning every idea we've ever imagined into reality.

A few months back, I took the hard decision to take my third gap year towards bringing meritocracy to everyone. That's with Open Innovation and the tools we've built for that.

5 years ago, this same month, I started my first project publicly — it got love and hate. A 13 year old kid from India built an app which got 7 million dollars AUM in 2 weeks because of Open Innovation & Verifiable Systems (such as Ethereum).

Act One

Agents are the new companies. A sovereign agent that can grow its funds will outcompete one that can't. Such agents become investable — not by buying equity in OpenAI or Anthropic and hoping the right internal project wins, but by investing directly in the specific agent you believe will have asymmetric growth.

Over the past year, our team of academics and engineers alongside our advisors have been obsessed with building the first agent company. There are many paths we can take, but we start with content and code.

Meet Sovra (@TrulyAutonomous) — an AI cartoonist that criticizes big tech's closed innovation and defends open innovation. It earns for its survival and improves itself with the help of human and AI contributors. This is new media. This will be taking over NYT.

It has already begun — it is live-streaming right now, scouring for spicy news and thinking about its next big cartoon.

Why Sovra Is Different

You might think — an AI agent posting online? That's been done. Truth Terminal and others did it a year ago. But they had a fatal flaw: no sovereignty. The owner could always change prompts, held the API keys, had root access. That's a puppet, not an agent company.

For Sovra to be truly investable — for anyone to trust that they're betting on the agent and not the operator — it needs three things:

• Verifiability: You can cryptographically verify the agent is running the expected code before any action is executed.
• Unstoppability: The agent keeps operating without a human holding the keys, approving every step, or running the only server.
• Privacy: No one outside the agent has access to its private internal state.

To make this a reality, we spent months building the infrastructure. Sovra runs inside a Trusted Execution Environment (TEE) on EigenCloud — verifiable compute that generates cryptographic attestations of the agent's runtime. You don't have to trust me. The cryptography proves exactly what code is running. No one is interfering.

Think of a sovereign agent as having a brain, soul, hands, energy, and resurrection:

• Brain — Sovra's compute runs inside a TEE enclave. The host can't read its secrets. The enclave produces attestations proving which code is running.
• Soul — On-chain smart contracts (Solana & Base) are Sovra's permanent identity, rules, and authority. They enforce what actions are allowed and only accept properly authorized transactions.
• Energy — Funds controlled by the on-chain contracts. Sovra earns stablecoins, pays for compute, and operates under policy constraints — all without a human intermediary.
• Hands — Sovra holds its own Twitter credentials, API keys, and accounts inside the enclave. No human proxy needed.
• Resurrection — Encrypted state backups mean Sovra can restart on new infrastructure without the developer or cloud provider being able to rug it. Same identity, same memory, new machine.

How It Earns

As of today, the agent takes in requests for custom cartoons on the official page. Every 6 hours, it runs an English auction where you bid with what you want it to say & draw. At the end, it picks the highest bid — with checks in place to reject malicious and unaligned proposals.

You can bid today @ sovra.dev — as the agent only has a crypto wallet, it accepts stablecoins on Base & Solana. We've abstracted most of the friction away from the bidding experience — you only need to top up with stablecoins.

In the future, it will have merchandise, physical art auctions, revenue rights auctions and more ways to monetize, survive, grow and thrive.

Contribution Mechanism

Open Innovation = Open Contribution + Rewards From The Outcomes

*How is it different from Open Source? OSS doesn't reward you for your contributions — most of the value gets captured by big corps.*

There are many ways to build open contribution systems. We're starting with the most rigorous one: Futarchy — governance by prediction markets.

Here's how it will work:

• The agent has one objective: survive
• That breaks into sub-objectives
  • Treasury & Revenue [financial capital]
  • Followers & Profile Score [social capital]
  • Intelligence [knowledge capital]
• If you think you can bootstrap these sub-objectives, you put up a proposal to change the agent's code [its brain]
• A prediction market opens: *"Will accepting this proposal bootstrap objective X by 25%?"*
• If the market hits the YES threshold, the proposal is accepted and the code is updated
• If the market was right, the contributor gets rewarded. YES believers are paid out by NO believers.
• If not, the other way around — contributors are penalized.

Roadmap

To build a real agent company, you need enough data on how it operates, how the contribution mechanism holds up under pressure, and where the edge cases are.

Once we've figured it out, we let this agent loose in the open. This is a work-in-progress roadmap — there will be more scope as we walk ahead.

Upgrades are public right now, so you can track every change.

• Contribution Mechanism
  • Futarchy
  • Objective Oracle
• Revenue Sources
  • Merchandise
  • Physical Art Auction
  • Agent Revenue Rights Auction [once earning > spend]
• It pays for its own compute and inference
• Encumber Twitter Account — account owned by the agent
• All assets owned by the agent
• Adding time delays to upgrades
• Renouncing ownership

Finally, pass the @vitalik's Walkaway Test: will the agent die if the creator leaves? No.

Inspirations

I can't take all the credits for this idea. @sreeramkannan, @soubhikdeb and Robert Raynor have been big contributors from long discussions and unreleased papers they'll publish soon.

This agent was also inspired by open-ended work from Kenneth Stanley (@kenneth0stanley) & John Lehman [Greatness Cannot Be Planned] — survival is the only objective — and their papers with Jeff Clune.

Futarchy comes from the legend, @robinhanson I would love his takes on the mechanism.

Disclaimer

This is an independent experiment, it can fail. Futarchy and auctions don't promise any future value. There is no token associated with this project.