About Provyn

The Problem

AI models generate legal briefs, financial analyses, medical recommendations, and business decisions every day. When those outputs are wrong — and they will be — the first question from regulators, courts, and insurers is: “Prove where this came from.”

Nobody can answer that question today.

What We're Building

Provyn is building two products to close the AI provenance gap:

Provyn Index Live

A public, searchable database of AI incidents and failures. Every entry is structured with actuarial-grade metadata — financial impact, litigation status, jurisdiction, root cause, and mitigation analysis. The data that risk analysts, insurers, lawyers, and compliance teams actually need.

Provyn SDK Coming Soon

A lightweight Python package that wraps your existing AI API calls and creates a cryptographically verifiable audit trail for every output. Model version, input hash, output hash, timestamp, human review status — chained and independently verifiable.

Why Independent Provenance Matters

Independent third-party provenance records carry more evidentiary weight than self-reported logs from AI providers — the same reason financial audits require external auditors. An AI company cannot credibly audit its own outputs. Provyn exists as an independent layer, across providers, with cryptographic guarantees that records haven't been tampered with.

Why Now

EU AI Act — Enforcement begins August 2026. Requires transparency obligations and audit trails for AI-generated content in high-risk applications.
ISO/IEC 42001 — Requires documentation of data provenance throughout the AI lifecycle.
Gartner Top 10 Strategic Technology Trends 2026 — Digital provenance named explicitly. Estimates organizations failing to invest by 2029 face sanction risks in the billions.

How the Index Is Different

The OECD AI Incidents Monitor and Partnership on AI's database catalog incidents. Provyn Index structures them for risk pricing. Every entry includes financial impact quantification, litigation tracking with outcome status, jurisdiction data, root cause analysis, mitigation analysis, and a timeline of how the incident evolved. These are the fields that make the data useful for insurance underwriting, liability assessment, and regulatory compliance — not just awareness.

Get in Touch

Building AI in a regulated industry? Working in insurance, risk management, or compliance? We want to hear from you.

Get early access to the Provyn SDK

Cryptographically verifiable audit trails for AI outputs. Coming soon.