NOAN is an AI knowledge system for entrepreneurs, and at its core is the "Fact Layer": it centrally manages a company's real, verified facts and provides them to agents, applications, websites, and teams through a unified API. The official emphasis is that "memory is what someone said, facts are what is true," so NOAN treats facts as approved, versioned, real-time served content rather than conversational memory.
Main capabilities:
- Fact Layer and API: Provides REST and MCP interfaces, and any model, agent, or application that supports REST or MCP can read the same set of facts. A single GET returns verified facts, with answers accompanied by source evidence.
- Fact versioning and traceability: Every fact has a complete version history, recording the owner, approver, and effective date; it supports viewing the state of the fact base at a given date; any answer can be traced back to which fact it came from and when it was true.
- Agent records: Each agent claims tasks, reports notes, and completes tasks, sharing the same task board with the team, and all actions are recorded in the activity timeline.
- Verity agent: Can join Slack channels, answer questions from verified facts, organize team conversations into cards pending verification, create tasks, contacts, notes, and reminders in NOAN, and run users' own agents.
- Unified management of business elements: Facts such as revenue, team, products, metrics, and strategy are unified into a "company brain"; notes, assets (templates and formats), and the contact network can all be accessed through the API.
- Real-time site rendering: Pricing pages and structured data can be rendered in real time from the fact layer, and changing one fact can synchronously update the page, API, and all agents; changelogs can also be written directly into the fact layer by agents.
- Agent Skill: Provides an open-source SKILL.md that can be used in more than 20 agents including Claude Code, Codex, Cursor, Copilot, etc., enabling agents to read verified facts.
- Rapid building: Uses pure REST and a bearer key, with no SDK or framework required; quickly build agents, internal applications, or websites that read facts.
How to use: Users can get started simply by entering a website URL; the NOAN API can be connected to any model; coding agents can also read the OpenAPI specification and key to build internal tools. The official team also provides a prompt that can be copied into Claude, ChatGPT, or another agent to have it read the API documentation and explain NOAN's use in its ownbusiness.
Use cases: Entrepreneurs and teams who want AI agents, applications, and websites to run on a unified factual foundation; need to reduce model guessing and token costs; need fact versioning, approval, and traceable answers; want to run company-level agents through Slack; and scenarios that need to unify CMS, CRM, and the context layer into the same fact system.
Limitations and missing information: The official text does not provide information such as specific pricing, plan details, a complete list of platform compatibility, data storage locations, or privacy policies; this content is not currently available.