Cognee

Cognee is an open source memory platform for AI agents. It gives agents persistent memory across sessions by combining graph, vector and relational retrieval, so an agent can recall what it did last week, connect facts that live in different systems, and follow domain rules it was never trained on. Developers install it as a Python package and call simple operations to remember, recall and improve memories, with citations preserved on answers. It serves three audiences. Teams connect sources such as Slack, GitHub, Linear, Notion and Google Drive to build a shared company knowledge base their agents can query. Agent builders add memory to their own agents through the SDK, an MCP server, or integrations with Claude Code and Codex. Enterprises use Cognee to generate ontologies that ground ambiguous business terms. The engine can run locally, in Docker, on premises or on Cognee Cloud, and the Enterprise tier adds bitemporal memory with conflict resolution, provenance on every answer, and personalization per user and per agent. Pricing is published. Free costs $0 per month and includes one workspace, 1M tokens, unlimited users and unlimited API calls. Standard is usage based at $1.00 per 1M tokens processed, plus $5 per month for each additional workspace, and adds data source integrations, code indexing for repositories and in app support. Enterprise is sold by contract as a bring your own cloud engagement with a dedicated support engineer, a Slack channel and a support SLA, offered in fixed engagement lengths, with a discounted Startup package. The open source engine remains free to run on your own infrastructure.