The AI GRC Accelerator for AI governance teams
A small team can run a real AI governance program.
AIGRCeler is a team of AI agents that learns about your AI systems from the people who own them, rates the risk, proposes controls, collects the proof and keeps watching. Your people make every decision.
Free and open source (Apache-2.0). Runs inside Claude Code on your Windows computer. Not legal advice and not a certification.
Who decided your AI was safe to use?
Every AI system needs an owner, a risk rating, controls and proof. AIGRCeler asks, drafts and records. Your people decide.
Want the full picture? Watch the 8-minute film, or see all our films on YouTube.
For the person who has to answer for the AI.
You may be the whole AI governance team. Your company uses AI in hiring, support, finance or code, and someone has asked: which systems, how risky, and where is the proof? The EU AI Act, the NIST AI RMF and ISO/IEC 42001 are on your reading list.
AIGRCeler does the legwork and leaves the judgment to you.
Ask about your AI
A 77-question discovery profile and an agent that asks the owner of each AI system the plain questions, so nothing hides in a spreadsheet.
Rate the risk, choose controls
A starter catalog of 104 controls in seven areas and 32 risk scenarios. Your company owns its copy and changes it by proposal.
Keep the proof alive
Evidence is fingerprinted when it arrives and expires on a date, or when the model version changes. Only verified controls earn credit against risk.
Four stages for every AI system.
Each system moves through the same path. The last stage never ends, because governing a live system never ends.
- Stage 1
Discover
What the system is, what it does, who it affects and what data it uses.
- Stage 2
Assess
How risky it is, and which controls it needs.
- Stage 3
Govern
AI Governance approves the controls and gives each one an owner.
- Stage 4
Operate
Owners send proof, the proof is checked, risk is re-scored, incidents are handled.
Operate never closes: proof expires, models change and incidents happen.
A person always decides.
The agents draft, ask, check, score and record. They never approve a control, accept a risk or start a stage on their own.
The agents
- Ask the plain questions and write to people in business words
- Draft records and propose controls
- Check evidence and score risk
- Hand every record to an engine that writes it, so an agent cannot hand-edit a status
Your people
- Approve every control
- Accept or reject every risk
- Start each stage
- Sign each decision in their own words, in a journal that cannot be changed without detection
Helps you prepare. Never says you comply.
AIGRCeler is built to speed up the work of getting ready for three frameworks. It maps its controls to them, so you can see where each control helps and keep the evidence in one place.
The map holds 65 ISO/IEC 42001 items, 32 EU AI Act items and 72 NIST AI RMF items. Each is mapped to a control, or to part of how the tool works.
- ISO/IEC 42001The management system standard for AI.
- EU AI ActThe European regulation on AI. Ask your lawyer what applies to your company.
- NIST AI RMF 1.0The US framework for managing AI risk.
Whether your company complies is for you, your auditors and your lawyers to decide. The map is a working tool, not legal advice.
See it before you trust it.
Brightwell Talent Partners is a fictional staffing company. We ran AIGRCeler on its hiring screener and published every record, from the first question to the dashboard.
You need Windows with PowerShell, Python 3 and Claude Code. The engine is plain Python and has no network code.
Want a hand?
The tool is free and works without us. If your team would like help setting it up or running it, we also offer consulting.