Getting Started

Everything you need to start building AI agents with persistent, structured memory. Follow the guides below to go from zero to production-ready in minutes.

What is Rekall?

Rekall is a memory platform for AI agents. It provides seven distinct memory types that mirror how biological memory works -- from short-lived session context to long-term consolidated knowledge that strengthens with use and fades with neglect.

Instead of treating every conversation as a blank slate, Rekall lets your agents remember past interactions, learn user preferences, build knowledge graphs, and maintain execution state across sessions. Memories are stored with rich metadata, semantic embeddings, and configurable decay curves.

Rekall works with any AI framework or model. Use the REST API directly, or integrate via first-class SDKs for TypeScript, Python, and Go. For Claude Desktop, Cursor, and other MCP-compatible tools, Rekall ships an MCP server with 40+ tools out of the box.

Why Memory Matters

person

Personalization

Agents remember user preferences, past decisions, and communication styles to deliver tailored experiences without re-prompting.

trending_up

Continuous Learning

Procedural memories let agents learn from successes and failures, refining their workflows over time without retraining.

groups

Shared Knowledge

Hive memory contexts let teams of agents share knowledge, coordinate, and build collective intelligence.

restart_alt

Resilient Execution

Execution memories provide checkpoint and resume capabilities so long-running agent tasks survive interruptions.

Memory types work together

Rekall's memory types are designed to complement each other. An agent might use short-term memory for the current conversation, episodic memory to recall past interactions, semantic memory to look up domain knowledge, and preferences to personalize its responses -- all in a single turn.

Get Started

Explore Further

Once you have completed the getting started guides, dive deeper into the seven memory types that power Rekall.

history
Episodic

Event-based memories capturing what happened, when, and in what context.

hub
Semantic

Knowledge graphs of entities, relationships, and facts.

account_tree
Procedural

Learned workflows, processes, and step-by-step procedures.

psychology
Long-Term

Consolidated knowledge with natural decay curves, strengthened by retrieval.

timer
Short-Term

Session-scoped working context powered by Redis for fast access.

play_circle
Execution

Agent task state with checkpoint, pause, and resume capabilities.

tune
Preferences

Learned user preferences inferred from interactions over time.

Rekall
rekall