Intent Record AB builds IRP, decision-provenance infrastructure for AI. We give teams and their agents a human-confirmed record of why decisions were made, so context that used to live only in people's heads survives every change of tool, model, and team.
A product you can use today, an open record format underneath, evidence that outlives the tools that made it. We are early, making accountability for AI decisions something teams can actually use.
Turn your decisions into audit-ready evidence for the EU AI Act, DORA, and beyond. Continuous, sovereign, and mapped to the obligations that actually apply to you.
An open, tool-agnostic decision-record format. Confirm a decision, it is written to a local, append-only record, and it stays forever.
Proof, not just a record. Verifiable proof of what was decided, by whom, and why, that anyone can confirm is genuine and unaltered without having to take our word for it.
The difference is neutrality and durability. A record is only worth something if a human stands behind it and it survives the systems that made it.
The reasoning behind a decision is usually lost the moment the tool, the model, or the team changes. These three steps keep it.
Your team decides, in the tools it already uses. A model or an agent may help. A human owns the outcome.
A named person confirms what was decided and why, in plain words. That human confirmation is what turns a log entry into a record someone can be accountable for.
The record is written to an append-only file you hold, and can be sealed with an independent timestamp, so anyone can check it existed and was not changed later, without taking our word for it.
Talk to us about IRP Compliance, the IRP protocol, or bringing decision provenance to your organisation.