Running compliance check…
Regulatory Risk Report
Multi-framework reasoning used as supporting evidence. Not a legal determination.
Metadata only — decision text is never stored.
Upload a CSV with columns: decision, category and any context columns. Returns a results CSV with risk flags, confidence, recommendation, and regulatory refs. Limit: 100 rows.
or drag & drop here
Upload a CSV of AI decisions with a demographic column. Pragma computes selection rates per group and flags any group below the EEOC 80% threshold — the same analysis required by NYC Local Law 144 annual bias audits.
Needs at least one demographic column and one outcome column · Up to 5,000 rows
applicant_id,race,gender,decision
1001,White,Female,advance
1002,Black,Male,decline
1003,Hispanic,Female,advance
...
One row per evaluation — SHA-256 input hash, no raw decision text stored. Proxy variables and firewall verdicts logged per ECOA & EU AI Act Art. 12 requirements. Hand this to a regulator.
Register your AI system for a complete 15-article EU AI Act compliance assessment covering Arts. 4, 5, 6, 9–15, 17, 25, 27, 30, and 33. Generates an auditable compliance certificate.
Run pre-built scenarios against the live Pragma engine. Each scenario mirrors a real AI system and shows which regulations pass or fail — the same evidence you'd hand a regulator.
Activate additional compliance frameworks beyond EU AI Act. Controls are tracked per-org and appear in audit reports.
EEOC 4/5ths rule — decision outcome rates across protected groups. Alerts fire when AIR < 0.8 for two consecutive weeks.
Create a workspace to share decision history with your team. Share the invite code with colleagues.
Use your API key to call Pragma from your own AI systems. Every decision your AI makes can be evaluated in real-time. The raw key is shown once — store it safely.
Configure rule severity, toggle jurisdictions, and add custom patterns for your org — no code deployment needed. Changes apply to all evaluations immediately.