Built by an engineer who got tired of shipping AI in the dark.
Every AI engineer knows the moment. You build an agent, it works in staging, you ship it, and then you have no idea what it is doing in production. No traces. No cost visibility. No way to know if a prompt change broke something quietly three weeks ago.
Ryva started as the tooling we wished existed. A framework that brings the same engineering discipline to AI systems that dbt brought to data pipelines and Terraform brought to infrastructure.
It is becoming an AI governance platform because that is what enterprise teams actually need. Not just testing, but the continuous compliance evidence that lets them defend every AI decision to regulators, lawyers, and boards.
We are building in public, open source first, and funded by the enterprise teams that need this most.
How we work
Open by default
Apache 2.0 licensed CLI, forever free. Enterprise features fund the open source work.
Honest about limitations
We document what Ryva does not do. We do not oversell.
Engineers first
Every product decision starts with the person running the terminal.
What we are building toward
2026
Full EU AI Act and Colorado AI Act coverage
Every article, every requirement, machine-readable evidence for regulators.
2027
The standard for AI governance in regulated industries
Ryva becomes the default compliance layer for AI teams in fintech, healthcare, and legal tech.
Beyond
The dbt of AI governance
The same engineering discipline that dbt brought to data pipelines and Terraform brought to infrastructure — applied to AI systems.
The full Ryva CLI is open source and Apache 2.0 licensed.
The CLI, the test runner, the lineage engine, the governance reporter — all of it is open source. Enterprise features like shared dashboards, SAML, and compliance exports fund the open source work.