# Mnemontron

> Mnemontron is provenance-aware knowledge infrastructure for AI agents and teams. It keeps the original evidence behind every answer, records who verified each claim, and returns citations that can be checked. It is pre-launch: there is no public API, hosted service or pricing yet.

All figures on this site are internal benchmarks run by the Mnemontron team in a development environment. The 100+ exabyte capacity figure is a projection, not a test that was run.

## Pages

- [Overview](https://mnemontron.com/index.md): Mnemontron keeps the original evidence behind every answer, records who verified each claim, and returns citations agents and people can check. Pre-launch; early access by request.
- [Benchmarks](https://mnemontron.com/benchmarks.md): Internal benchmark results for Mnemontron retrieval, abstention, storage repair and restore, with the setting of each measurement and what has not been measured.
- [Security](https://mnemontron.com/security.md): How Mnemontron protects tenant data: passkey-only sign-in with step-up re-authentication, per-tenant AES-256-GCM envelope encryption, row-level tenant isolation, a hash-chained audit log and signed receipts.
- [FAQ](https://mnemontron.com/faq.md): Answers to common questions about Mnemontron: availability, pricing, how citations and verification work, agent integrations and data security.

## Optional

- [Full text of all pages](https://mnemontron.com/llms-full.txt)