Skip to main content

​
What is MemoryLake?

MemoryLake is a memory platform for AI applications. You feed it documents and conversations, and it automatically extracts structured facts that your agents and applications can search and retrieve. Think of it as long-term memory for AI — persistent, queryable, and organized by who said what. MemoryLake is built around three pillars:

Document Memory

Import files into a project. MemoryLake indexes them for semantic search so your AI can find relevant content across your entire knowledge base.

Conversation Memory

Submit user-assistant conversations. MemoryLake automatically extracts structured facts — preferences, decisions, context — and associates them with the right actors.

Unified Search

Query across both documents and facts with a single natural language search. Get the most relevant information regardless of how it entered the system.

​
Resource Model

MemoryLake organizes data in a clear hierarchy:
Workspaces provide data isolation — like tenants. Each workspace has its own actors, projects, and search index. Actors represent participants. Every human user or AI assistant gets an actor identity. Facts about an actor follow them across all projects in the workspace. Projects are knowledge containers. A project holds documents, conversations, and the facts extracted from them. You typically create one project per use case, domain, or application.

​
How It Works

1

Ingest

You bring data into a project through two channels: import documents from the Library, or submit conversations via the API. Both channels feed the same project.
2

Extract

MemoryLake processes your content automatically. Documents are indexed for semantic search. Conversations are analyzed to extract structured facts — things like user preferences, stated goals, decisions, and context.
3

Search

Query the workspace with natural language. MemoryLake searches across both document content and extracted facts, returning the most relevant results in a single response.

​
Working in the Console

The API is one way in; the console is the other. Everything above is also available as a product surface:

Console Walkthrough

Chat with memory in the Playground, upload files, and review what was extracted — no code.

Document Management

The Library: folder hierarchies, uploads, comments, and supported formats.

Memories

Browse, search, trace, and resolve conflicts in extracted memories.

Workspaces & Projects (console)

Create projects, associate documents, and navigate the project dashboard.

External Connectors

Mount WPS and Lark document libraries into your personal library.

Integrations

MCP over OAuth2, the OpenClaw agent plugin, and REST API access.

​
Scenarios

Personal AI memory

One private memory shared by every AI tool you use.

Team knowledge base

Governed organizational memory that outlives personnel changes.

Memory for coding agents

Auto-recall, auto-capture, and auto-upload for OpenClaw agents.

Build apps with memory

Per-user long-term memory in your own product.

​
Get Started

Quick Start

Create a workspace, submit a conversation, and search your memories in 5 minutes.

Workspaces & Projects

Understand the resource hierarchy and when to use multiple workspaces vs. multiple projects.

Actors & Memory

Learn how actors provide per-user memory that follows users across projects.

Memory Pipeline

Understand how conversations become structured facts and how search unifies everything.