The three things audit and compliance ask, that today nobody can answer.
“Show me the controls applied to this agent decision.”
Every Vortalis decision carries the policy that was applied, the inputs the policy saw, and the outcome. One record per decision. Searchable. Exportable. Versioned.
“Prove this evidence has not been altered.”
Audit records are SHA-256 hash-chained, and exports and evidence packs are signed. Your auditor verifies the chain and the signature offline using a command-line utility we provide that runs locally. They need a copy of our public key set; once they hold it, verification does not depend on Vortalis being available, and they do not need a contract with us to verify. Rows written since the last checkpoint (hourly by default) rest on the hash chain alone.
“Map your controls to my framework.”
We maintain the mapping between Vortalis controls and the obligations under the major UK and EU frameworks for AI in financial services and other regulated sectors, including AI-specific and information-security standards. Mapping is delivered as evidence in a structured pack your auditor recognises.
What you receive.
- An evidence pack per decision, per agent, or per population, on request.
- Offline verification tooling for your auditor.
- A regulatory coverage map covering the frameworks your supervisor cares about.
- Audit export controls (JSON and CSV) and one platform-wide audit retention period, which is not configurable per customer.
- A documented control narrative your auditor can attach to a working paper.
- A reconstructable delegation trail: every action taken under an explicit delegation grant is checked against policy and recorded in the hash-chained audit trail, so you can show exactly which agent authorised which downstream action after the fact. Delegation and joint-action checks are built as options, off by default and enabled per tenant. Where joint-action evaluation is on, a rule can also match a syntactic, time-windowed pattern over audited event text. There is no semantic understanding of what agents say to each other, and this does not solve emergent multi-agent coordination in the general case.
- Intent-vs-action drift alerts your MRM team can point to, measuring the divergence between a session's declared intent and what it actually did. A bounded telemetry signal, not a claim to have solved goal-drift as a research problem.
On the regulatory coverage map.
The full mapping is shared under a sales conversation rather than published openly. This is for two reasons. First, the mapping is current and we want to keep it current for you specifically rather than have a static page age in public. Second, the crosswalk represents real work and real IP, and we share it with serious buyers.
Get the regulatory coverage mapRelated reading: AGAP® audit and conformance · EU AI Act mapping
Show this to your audit lead first.
Bring an actual control you need to evidence, or a finding from your last audit you do not want to repeat. We will show how Vortalis would have produced the evidence, in a format your auditor recognises.
See a signed evidence pack