Launching today
Fact0
The universal fact layer for AI agents
20 followers
The universal fact layer for AI agents
20 followers
Log every autonomous action, tool call, and human approval in a tamper-evident cryptographic chain. Pass enterprise security reviews in days, not months.
Hey 👋
I’m Yash, the creator of Fact0.
What inspired it / The Problem:
Autonomous AI agents are incredibly powerful, but their non-deterministic nature makes them a trust and compliance nightmare when selling to enterprises. Whenever B2B agent startups go through security reviews, CISOs and compliance auditors ask: "Can you prove exactly what your agent did, which LLM calls it made, who approved its actions, and that these logs haven't been retroactively modified?"
Building a secure, tamper-evident audit trail and visual execution replay system is a massive undertaking for early-stage teams. I built Fact0 to be the universal fact layer for AI agents, giving developers a plug-and-play compliance infrastructure.
How Fact0 Helps:
• Tamper-Evident Hash Chains: Every event, tool call, and decision is cryptographically chained using SHA-256 at the database layer. If any log row is edited or deleted, the chain breaks immediately.
• Visual Execution Replay: We reconstruct the directed acyclic graph (DAG) of your agent runs so you can replay execution frames step-by-step and debug what went wrong.
• Compliance-Ready Exports: One-click SOC 2-ready evidence packs (PDF/ZIP exports) and secure, time-limited read-only share links for external auditors.
• Fact0 Spotlight: An in-dashboard AI assistant that lets you ask questions about your audit logs, failed runs, or compliance status in plain English.
• Developer-First & OTel Native: Lightweight Python, TypeScript, and Go SDKs with async batching so logging never slows down your agents. You can also ingest logs directly via OpenTelemetry.
How the Approach Evolved:
Fact0 started as a core Go API and a simple PostgreSQL ledger to solve the cryptographic verification problem. However, as I spoke with developers, I realized that compliance is only half the battle. The other half is explaining agent decisions to non-technical stakeholders. This led to focusing heavily on interactive visual timelines and audit-sharing portals, making it as easy to prove agent behavior to customers and compliance teams as it is to debug it.
Fact0 is live at fact0.io, and we'd love to hear your thoughts.
What frameworks are you building agents with? How are you handling compliance today?
Looking forward to your feedback and questions! 🚀