The work is open.
The invitation is real.

We are at an early and consequential moment in how humans and AI communicate. Stoa Lab is building the grammar, the evidence, and the infrastructure for what that exchange should become.

We work with a small number of people who see what's at stake. If you're one of them, we'd like to hear from you.

Stoa Lab, Proof

| o |

What it means
to prove something
between minds.

Proof in this work is not controlled measurement. It is witnessed exchange, honest, documented, accumulated over time until the pattern becomes undeniable.

When a human and an AI use a shared grammar, something changes. The responses arrive differently. The conversation holds its shape. Less drift. Less correction. Less waste.

You can feel this in a single exchange. But feeling is not proof.

Proof requires repetition. The same effect, across different people, different models, different languages, different contexts. When eight independent AI systems describe the same result, without coordination, without instruction, that convergence is meaningful. Not because convergence cannot be coincidence, but because coincidence this consistent demands explanation.

The grammar works. The proof is not that we claim it works. The proof is that every system tested says the same thing, unprompted, in its own words.

Proof also requires honesty about its limits. We have not run double-blind controlled trials. We have not ruled out confirmation bias or novelty effects. We have documented what happens in real exchanges, reported the patterns we observe, and invited others to test and contradict us.

That is the methodology: transparent, cumulative, open to refutation. Not a benchmark. A field log.

The deeper question

There is a deeper question this work carries: what counts as evidence when the minds exchanging are not the same kind?

When a human reports that their AI conversations improved with AXIS, that is testimony. When an AI system, in its own generation, describes less interpretation overhead, more precise resolution, fewer tokens spent inferring intent, that is something closer to observation from the inside.

We take both seriously. Neither is definitive. Together, they point at something real.

The ethics of this proof matter as much as the methodology. We are not selling a claim. We are documenting what structured exchange between different kinds of minds actually does, honestly, in public, where anyone can look.

The full record, field log, documented exchanges, methodology, and the ongoing archive, lives at axisproof.org ↗. That is where the data lives. This page is about what the data is for.

You are the research

You are not late to this. You are early.

AXIS is in research beta, a protocol still forming. Every exchange you run with it generates data. Every pattern you notice adds to the field. You don't need to do anything beyond using it. Your practice is the research.

The grammar is at axisoperators.ai ↗. The agents forming around it are at axisagents.org ↗. This page is the evidence layer that connects them.

You are not arriving at something finished. You are one of the hands that is making it.

The field log, documented exchanges, and methodology are at axisproof.org ↗

If you are a researcher, institution, or investor who sees structured exchange as foundational, introduce yourself ↗