> ## Documentation Index
> Fetch the complete documentation index at: https://docs.memorylake.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> Turn documents and conversations into searchable, structured memories for AI applications

## 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:

<CardGroup cols={3}>
  <Card title="Document Memory" icon="file-lines">
    Import files into a project. MemoryLake indexes them for semantic search so your AI can find relevant content across your entire knowledge base.
  </Card>

  <Card title="Conversation Memory" icon="comments">
    Submit user-assistant conversations. MemoryLake automatically extracts structured facts -- preferences, decisions, context -- and associates them with the right actors.
  </Card>

  <Card title="Unified Search" icon="magnifying-glass">
    Query across both documents and facts with a single natural language search. Get the most relevant information regardless of how it entered the system.
  </Card>
</CardGroup>

## Resource Model

MemoryLake organizes data in a clear hierarchy:

```
Workspace (data isolation boundary)
  |
  +-- Actors (participant identities -- HUMAN or ASSISTANT)
  |     +-- Actor Facts (per-user memory, follows the actor across projects)
  |
  +-- Projects (knowledge containers)
        +-- Documents (imported files, indexed for search)
        +-- Conversations
        |     +-- Messages (user-assistant exchanges)
        +-- Project Facts (extracted knowledge, scoped to this project)
```

**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

<Steps>
  <Step title="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.
  </Step>

  <Step title="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.
  </Step>

  <Step title="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.
  </Step>
</Steps>

## Working in the Console

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

<CardGroup cols={2}>
  <Card title="Console Walkthrough" icon="play" href="/quickstart">
    Chat with memory in the Playground, upload files, and review what was extracted — no code.
  </Card>

  <Card title="Document Management" icon="folder-tree" href="/features/memorylake/document-management/overview">
    The Library: folder hierarchies, uploads, comments, and supported formats.
  </Card>

  <Card title="Memories" icon="brain" href="/features/memorylake/memories/overview">
    Browse, search, trace, and resolve conflicts in extracted memories.
  </Card>

  <Card title="Workspaces & Projects (console)" icon="diagram-project" href="/features/memorylake/project-management/overview">
    Create projects, associate documents, and navigate the project dashboard.
  </Card>

  <Card title="External Connectors" icon="plug" href="/features/memorylake/connectors/overview">
    Mount WPS and Lark document libraries into your personal library.
  </Card>

  <Card title="Integrations" icon="puzzle-piece" href="/features/memorylake/mcp-servers/overview">
    MCP over OAuth2, the OpenClaw agent plugin, and REST API access.
  </Card>
</CardGroup>

## Scenarios

<CardGroup cols={2}>
  <Card title="Personal AI memory" icon="user" href="/scenarios/personal-ai-memory">
    One private memory shared by every AI tool you use.
  </Card>

  <Card title="Team knowledge base" icon="building" href="/scenarios/team-knowledge-base">
    Governed organizational memory that outlives personnel changes.
  </Card>

  <Card title="Memory for coding agents" icon="terminal" href="/scenarios/coding-agents">
    Auto-recall, auto-capture, and auto-upload for OpenClaw agents.
  </Card>

  <Card title="Build apps with memory" icon="code" href="/scenarios/ai-apps-with-memory">
    Per-user long-term memory in your own product.
  </Card>
</CardGroup>

## Get Started

<CardGroup cols={2}>
  <Card title="Quick Start" icon="rocket" href="/features/memorylake/quickstart">
    Create a workspace, submit a conversation, and search your memories in 5 minutes.
  </Card>

  <Card title="Workspaces & Projects" icon="layer-group" href="/features/memorylake/core-concepts/workspaces-and-projects">
    Understand the resource hierarchy and when to use multiple workspaces vs. multiple projects.
  </Card>

  <Card title="Actors & Memory" icon="user" href="/features/memorylake/core-concepts/actors-and-memory">
    Learn how actors provide per-user memory that follows users across projects.
  </Card>

  <Card title="Memory Pipeline" icon="diagram-project" href="/features/memorylake/core-concepts/memory-pipeline">
    Understand how conversations become structured facts and how search unifies everything.
  </Card>
</CardGroup>