Cognee
Open-source memory engine for AI agents, with a hosted option.
- Tool type
- Open-source AI memory engine
- Made for
- Developers and teams building agents
- Runs on
- Local machine, own infrastructure, or Cognee Cloud
- Open-source engine runs locally or self-hosted; Cloud is hosted; the experimental Rust SDK targets on-device builds.
What it does
Cognee is an open-source memory engine for AI agents. It builds a knowledge graph that you fill with remember and query with recall. Sources such as Slack, GitHub, Linear, warehouses, docs, and APIs can be connected so agents retrieve what a team already knows, with citations kept attached to answers.
It is used through a Python library, a CLI, an MCP server, an HTTP API, and direct integrations with agents such as Claude Code, Cursor, LangGraph, and OpenClaw. Custom ontologies and domain rules guide how agents interpret terms, and feedback can update recall weights for later runs.
The same engine runs locally or on your own infrastructure, or as Cognee Cloud, a managed service with a web UI and dataset-level permissions. Enterprise deployments are delivered as bring-your-own-cloud engagements. The Rust SDK targets on-device pipelines and is experimental, and a TypeScript SDK serves Node.js.
What it can do
- Builds knowledge graphs from documents and connected data sources via remember
- Retrieves cited answers from stored memory via recall
- Generates and applies custom ontologies and domain rules for agents
- Stores agent experience so agents can recall past work across sessions
- Accepts feedback to adjust recall weighting for future runs
- Provides an MCP server so compatible agents read and write memory
Works with
- Claude Code
- Codex
- Cursor
- LangGraph
- OpenClaw
- Hermes Agent
- Slack
- GitHub
- Linear
- Notion
Before you choose
- Rust SDK is experimental and built from source with Cargo, not via pip.
- Standard Cloud pricing is usage-based per token plus per-workspace fees.
- Enterprise deployments are custom BYOC engagements with pricing not published.
Listed under
Get started
Run pip install cognee, then follow the quickstart; Cloud sign-up is also offered.
- Install the open-source package with pip install cognee.
- Run the quickstart to build a first memory example.
- Connect an agent through the MCP server or SDK, and add sources such as GitHub or Slack.
Plans and prices
Prices in USDThe Free plan includes 1 workspace and 1M tokens with no card required. Standard is priced at $1.00 per 1M tokens processed plus $5 per additional workspace per month. Enterprise is a custom BYOC engagement; contact sales for pricing.
- 1 workspace and 1M tokens included
- Unlimited users and API calls
- Unlimited workspaces
- Slack, Notion, Linear, Google Drive integrations
- BYOC deployment in your own cloud
- Dedicated support engineer and SLA
Source: cognee.ai/pricing, checked 29 Oct 2026. Prices can change; check before you buy.
Sources for this overview
Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.
- Cognee - Open-Source Agent Memory Platform ↗www.cognee.ai
- Build AI Memory with Cognee - Cognee Documentation ↗docs.cognee.ai
- Pricing | Cognee - AI Memory Engine Plans ↗www.cognee.ai
Ratings and reviews
No ratings for Cognee yet.
What’s changed
Free accountSignals
Free accountData and security
Free accountCompare with similar tools
Free accountWhere this comes from
How we research- Company website
- cognee.ai: product, pricing and policy pages
- YC directory mirror
- Batch and company name
- Web search
- Confirms the official site
- Social accounts
- Accounts the company links, and its posts on X
- Human review
- Checked by a person
- knowledge_base
- manual_review
Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.