Pratvi AI
Private beta · invite required

Ship AI features regulatorswon't block.

The governance OS for compliance, risk, and engineering teams in healthcare, finance, government, and other regulated industries. Inventory every model, prove every decision, satisfy every framework — from one platform.

Talk to founder Try the live demo
Pratvi AI dashboard — sample interface for illustration. Values shown are synthetic, not customer data.
audit · sha-256 chain 
The challenge

The compliance deadline is real. The exposure is now.

Regulators have set the date and the penalties. Most teams are still governing AI with spreadsheets and good intentions.

Aug 2026EU AI Act high-risk obligations enforce

High-risk system duties under the EU AI Act apply from August 2026.

€35Mor 7% of global turnover in fines

Maximum penalty for prohibited-practice breaches — EU AI Act Art. 99.

11–20 hrsper week lost to manual governance

Industry observation: time sunk into spreadsheet-based AI oversight.

No trailMost regulated AI runs ungoverned

Most regulated AI today has no inventory, no audit trail, no evidence.

Every ungoverned model is regulatory, reputational, and operational exposure. Pratvi closes the gap.

One platform spanning the surface area of regulated AI — from foundation models to vertical-specific obligations.

12
Platform modules
19
Regulated verticals
32
Frameworks supported

Built to support

Every regulator your AI will face.

One platform. 32 frameworks across six regulatory domains — all mapped to capabilities that exist in the codebase today.

AI-specific

  • EU AI Act
  • NIST AI RMF 1.0
  • ISO/IEC 42001
  • OMB M-24-10
  • Colorado AI Act
  • Texas TRAIGA

Healthcare

  • HIPAA
  • HITRUST CSF
  • CMS-0057-F
  • FDA AI/ML SaMD
  • 21 CFR Part 11
  • FHIR R4 AuditEvent

Financial services

  • NAIC Model Bulletin
  • SR 11-7
  • OCC 2011-12
  • ECOA
  • FCRA
  • GLBA Safeguards
  • BSA/AML
  • HMDA
  • Colorado SB21-169

Privacy & Data Protection

  • GDPR
  • FERPA
  • CCPA / CPRA
  • UK GDPR

Security & Infrastructure

  • ISO/IEC 27001
  • SOC 2
  • CMMC 2.0
  • NERC CIP

Sectoral & Other

  • ITAR
  • DFARS 7012
  • NYC Local Law 144
See every framework and what we support →
Why Pratvi

Generic governance tools weren't built for the regulated stack.

Most AI governance products bolt fairness dashboards onto a model registry. Pratvi was designed from day one around the obligations that actually break in regulated industries — vertical depth, examiner-grade evidence, and per-decision explainability.

Built for regulators, not buzzwords
Every framework on this site traces to a service in the codebase. EU AI Act Article 9 risk management. SR 11-7 conceptual soundness. ECOA Reg B 30-day adverse-action windows.
Vertical depth, not one-size-fits-all
19 verticals with 85 industry-specific sub-modules — prior-auth turnaround tracking for health insurance, NAIC Model Bulletin coverage for P&C, NERC CIP-013 for energy.
Evidence that holds up in an exam
SHA-256 hash-chained audit trail. FHIR R4 AuditEvent export. One-click evidence packs across 14 frameworks. Per-decision explanations in patient, clinical, and regulator-grade language.

12 modules. One platform. Continuous coverage.

Each module solves a specific governance obligation that today is filled with spreadsheets, point tools, or nothing at all. All 12 are built into one platform — your audit trail, bias monitor, and regulatory mapping speak the same language.

01

AI Model Inventory

One source of truth for every AI system in your organization.

02

Compliance Engine

Map every model to every framework that governs it.

03

Immutable Audit Trail

SHA-256 hash-chained logs of every AI decision.

04

Bias & Fairness Monitor

Catch fairness regressions before they become violations.

05

Drift & Performance Monitor

Statistical detection when your models stop working as designed.

06

Security Posture (MITRE ATLAS)

Threat-model your AI systems against the actual attack surface.

07

Explainability Engine

Per-decision explanations that hold up in front of regulators.

08

Confidence & Verification

Calibrated confidence, multi-model verification, automatic escalation.

09

Adverse Action Notice Engine

Automatic notices that satisfy ECOA, FCRA, and ACA §1557.

10

Evidence & Incident Workbench

Examiner-ready evidence packs and regulator-deadline-aware incident playbooks.

11

Executive Dashboard

Board-ready governance scores, narratives, and roll-ups.

12

Agentic Governance Layer

Autonomous compliance, governance, and reporting agents — with MCP integrations and scheduled jobs.

Explore each module in depth

19 regulated verticals. 85 vertical-specific sub-modules.

Pratvi is built for the depth of vertical-specific regulation that actually matters in your industry — not a generic 'enterprise AI' overlay. Each vertical has 2–12 industry-specific compliance modules underneath.

Health Insurance

Governance for payers running AI on member decisions.

12 vertical-specific sub-modules

HIPAA
CMS-0057-F
HITRUST

Property & Casualty Insurance

Underwriting and claims AI under NAIC scrutiny.

6 vertical-specific sub-modules

NAIC Model Bulletin
State DOI

Life Insurance

Risk classification AI without disparate-impact exposure.

3 vertical-specific sub-modules

NAIC Model Bulletin
State DOI

Banking

Model risk management for credit, fraud, and BSA/AML.

4 vertical-specific sub-modules

SR 11-7
ECOA
GLBA

Credit Unions

NCUA-aligned governance for member-facing AI.

2 vertical-specific sub-modules

NCUA
ECOA

Mortgage Lending

Fair-lending AI under HMDA, ECOA, and CFPB lens.

3 vertical-specific sub-modules

HMDA
ECOA
HUD
See all 19 verticals

Real obligations. Concrete workflows.

Hypothetical scenarios that show how the platform's modules address common regulatory concerns. Each scenario maps to capabilities that exist in the codebase today.

ACA Section 1557
NAIC Model Bulletin

Bias monitoring for utilization-management AI

A health insurer running AI-assisted prior-authorization decisions needs to demonstrate no disparate impact on protected classes — a core concern under ACA Section 1557 and the NAIC Model Bulletin. Without continuous monitoring, a regression in the underlying model can introduce statistically significant disparities long before a manual review would surface them.

SR 11-7
OCC 2011-12

SR 11-7 model validation for credit-decisioning AI

A bank deploying machine-learning credit models is subject to SR 11-7 model risk management. Each model needs documented conceptual soundness, ongoing monitoring, outcomes analysis, and an independent validation trail.

EU AI Act
ISO 42001

EU AI Act high-risk system conformity assessment

An organization deploying AI in employment, education, law enforcement, or essential services across the EU must classify each system under EU AI Act Annex III, complete a conformity assessment, and maintain Article 14 human-oversight controls.

More use cases
Trust

Built for regulated data — from day one.

Encryption everywhere

AES-256-GCM at rest, TLS 1.3 in transit, application-layer field encryption for sensitive data.

Tamper-evident audit

SHA-256 hash-chained logs satisfy 21 CFR Part 11, SR 11-7 retention, and HIPAA §164.312(b).

Tenant isolation

Row-level security, per-tenant encryption keys, no shared compute pools.

MFA supported

Multi-factor authentication available for all accounts, with org-wide enforcement controls. Hardware-key support planned.

Targeting SOC 2 Type II

Audit targeted post-GA + 12 months. HITRUST CSF r2 targeted post-GA + 24 months.

Sub-processors disclosed

Full sub-processor list published at /legal/subprocessors. GDPR Article 28 compliant.

Read the trust center →

Ready to see Pratvi AI?

We're in private beta. Talk to the founder and we'll set up a 30-minute walkthrough mapped to your specific compliance obligations.

Talk to founder Explore the platform

Currently in private beta. We onboard customers manually. No demo data uploaded.