About ArkForge
We believe that as AI agents become autonomous economic actors, every transaction they make should be independently verifiable.
AI agents are increasingly making decisions, calling APIs, and processing payments on behalf of humans. But when something goes wrong — a disputed charge, a failed delivery, a compliance question — there is often no verifiable record of what actually happened.
ArkForge exists to solve this problem. We build infrastructure that generates cryptographic proofs for every agent transaction, timestamped and logged by third parties independent of ArkForge.
Our goal is simple: no promises, only proofs.
Trust Layer is our flagship product: a certifying proxy that sits between an AI agent and any target API. For every transaction, it:
Every proof can be checked against the public ArkForge Proof Specification. The system produces a Proof Triptych: machine-readable JSON (L1), HTTP headers (L2), and a human-readable HTML page (L3).
Every call produces a cryptographic proof, and its timestamp and public log entry can be checked without us.
Our code is MIT-licensed. Our specification is CC BY 4.0. Inspect, audit, and contribute.
Hosted in France. GDPR-native. Your data stays in the European Union.
Once its batch is anchored, altering a proof breaks its timestamp and its public log entry. Informational fields such as dispute status can still be updated.
ArkForge is a French company building verifiable infrastructure for the AI economy. We are focused exclusively on Trust Layer and the challenge of making autonomous agent transactions provable and auditable.
We run our own automated operations through Trust Layer: outbound actions taken by our agents go through the proxy and receive a proof, like any customer call.
Explore Trust Layer or get in touch.