Why Curos exists

Artificial intelligence is quickly becoming infrastructure — the kind of technology that shapes how people work, learn, get answers, and make decisions. Infrastructure of that scale should be accountable to the public interest, the way roads and libraries are. We believe it should not be accountable to a quarterly earnings call.

That is the hole Curos is built to fill. We are an independent non-profit AI lab. Our product, Ichnus, is free for everyone. Our research is published in the open. Our decisions are governed by a charter, an independent board, and the people who use what we build.

What we believe

Four commitments run through everything Curos does:

  • Non-profit by structure. No shareholders, no investors demanding growth at any cost. Every decision is measured against our mission.
  • Free by default. Ichnus is free to use, because good AI shouldn't be gated behind a paywall for the people who need it most.
  • Open about how it works. We publish what we can about how our models are trained, tested, and deployed. Trust is earned in the open.
  • Accountable by design. We publish our safety evaluations and give the people who use Ichnus a real say in how it behaves.

What we do

Curos does two things that feed each other. First, we do research — on interpretability, alignment, model evaluations, reasoning, multi-agent systems, and the efficiency that makes all of it affordable. Second, we build a product people can use today: Ichnus, powered by three model tiers.

How we got here

Curos is young. Here is the short version of how it went from an argument to an organization with a product people actually use.

Feb 2025

Curos comes out of stealth

Founded by a small group of researchers and engineers who left better-funded labs to build something structurally different: an AI organization with no shareholders.

Mar 2025

The founding charter is adopted

Non-distribution, free access, openness, and independent oversight are written into a charter the board cannot quietly amend. See the full governance structure.

May 2025

Ichnus launches, free from day one

The first public release, powered by Krus. No waitlist, no paid tier gating basic capability — a stance we have not moved from since.

Oct 2025

First public safety evaluation report

Curos publishes what it tested, what it found, and what it didn't test — the format every release report has followed since. Read our approach to evaluations.

Mar 2026

First annual transparency report

Funding sources, spending, and salary bands, published in full. The transparency page is updated on the same cadence every year since.

Jul 2026

Corus ships, free with fair-use limits

The frontier tier of the Ichnus family — Curos's answer to the idea that the most capable models have to be paywalled.

You can read about the research agenda on the research page, our approach to safety and evaluation on the safety page, and exactly where our money comes from on the transparency page.

Who builds Curos

We are a small team of researchers, engineers, and people who believe AI should be held to a public standard. Most of us work remotely; none of us is motivated by equity in a company that could be sold.

leadership

Amara Okafor

Co-founder & Executive Director

Leads Curos and carries its mission: AI that answers to people, not profit. Former AI systems researcher who spent years inside commercial labs before helping found Curos.

leadership

Lucas Ferrand

Co-founder & Head of Research

Sets the research agenda across interpretability, alignment, evaluations, reasoning, multi-agent systems, and efficiency. Believes trust is earned in the open.

research

Priya Raman

Head of Alignment

Leads the alignment team: how Ichnus models learn to be helpful, honest, and honest about what they cannot do.

research

Tomás Herrera

Head of Evaluations

Builds and owns the evaluation suite behind every Ichnus release, from capability benchmarks to adversarial red-teaming.

research

Jonas Weber

Interpretability Research Lead

Studies the internal structure of our models — what circuits do what, and whether we can point at the mechanism behind a behavior.

research

Maya Lindgren

Senior Researcher, Reasoning

Works on compositional and step-by-step reasoning, and on models that say "I don't know" more often than they bluff.

research

Sasha Petrova

Senior Researcher, Multi-agent Systems

Studies how collections of models coordinate, verify each other, and stay honest when more than one agent is in the room.

engineering

Eli Nakamura

Research Engineer, Efficiency

Makes frontier capabilities affordable and fast — routing, quantization, and serving systems that keep Ichnus free to run.

engineering

Hana Sato

Engineering Lead, Ichnus Runtime

Owns the product that millions of people touch every day: the assistant, its apps, and the systems that keep it up.

policy

Grace Adeyemi

Director of Policy & Governance

Translates research into commitments. Runs the oversight bodies, writes our charters, and keeps the org honest to them.

ops

Sam Whitfield

Head of Operations & Funding

Keeps Curos financially sound and independent — grants, donations, and a lean budget, all reported in the open.

community

Daniela Rossi

Head of Community & Communications

Connects the people who use Ichnus to the people who build it, and publishes the transparency reports.

A note on what this site is

Curos and Ichnus are a concept project — a personal exercise in design and writing, not a real organization. None of the people, products, papers, or figures on this site are real. If you are here because you expected a working assistant, you will not find one. We hope the site is useful as an example of what accountable, transparent AI infrastructure could look like.

← Back to home