Mnemos AI
Documentation

Build, govern, and operate on Mnemos AI

Mnemos is the organizational memory platform: it captures how your company actually works, governs that knowledge with role-aware permissions, and makes it operational through AI. These docs cover every surface — product, admin, integrations, and the API.

Start here

New to Mnemos? Spend ten minutes with the quickstart guide. It walks you from workspace creation to your first AI-cited answer.

How the docs are organized

Each top-level section in the sidebar maps to one of five lenses on Mnemos. The product sections describe behavior; the administration sections describe controls; the developer sections describe contracts you can build against. Within each, pages move from concepts to recipes to references.

  • Concept pages explain a model — what it is, why it exists, and what it doesn't do.
  • Guides walk through a real task end to end with checkpoints.
  • References enumerate fields, errors, rate limits, and event payloads.

Get started

From zero to your first AI interview in under ten minutes. Concepts, quickstart, and how Mnemos is organized.

Open quickstart

Product

How the interview agent, memory graph, AI search, SOP library, and continuity engine fit together.

Tour the product

Administration

Members and roles, SSO, SCIM provisioning, API keys, audit log, and billing controls.

Admin overview

Integrations

Connect Slack, Notion, Google Workspace, Microsoft 365, Salesforce, and your own services via webhooks.

Browse integrations

Developer reference

REST API, SDKs, error model, rate limits, and per-endpoint contracts for every resource Mnemos exposes.

Read the API reference

Memory graph

Entities, relationships, and the ontology that turns transcripts into a queryable map of how your company works.

Explore the graph

What Mnemos is

Mnemos is not a wiki. It is a system of record for operational knowledge: the "how" behind every workflow, decision, and edge case in your company. An AI agent conducts voice-led interviews with your team, extracts entities and SOPs into a permission-aware memory graph, and serves that knowledge back as cited answers, ramped onboarding journeys, and continuity-risk dashboards. Every read and every write is auditable; every AI answer carries lineage.

The platform is tenant-isolated by row-level security in Neon Postgres, wraps LLM calls through a shared inference layer, and runs across nine domain services behind a FastAPI edge. The docs you are reading are generated from the live api-gateway OpenAPI specification.

Looking for a concept rather than a step? Read the concepts overview for definitions of workspaces, organizations, sessions, SOPs, and the memory graph.