Pieces
Free tierInfinite AI memory for developers — automatically captures everything you work on, across every app.
Free tier available·Technical·Powered by Multiple (cloud and local providers, bring-your-own-key supported)·API available
Key strengths
OS-level memory capture across all apps without manual effortLTM-2 Long-Term Memory Engine for persistent, time-aware contextPrivate by default — fully on-device with no cloud dependencyMCP integration with GitHub Copilot, Claude, Cursor, and GoosePlugin ecosystem spanning Chrome, VS Code, and other developer tools
Free tier + paid plans
US
Self-hostable
No ratings yet
Developer Integration & API Usage
MCP (Model Context Protocol)
Pieces exposes an MCP interface that connects your personal long-term memory to supported LLMs and AI coding assistants. Compatible hosts include GitHub Copilot, Claude, Cursor, and Goose. This allows the model to query your Pieces memory at inference time, grounding responses in your actual work history.
Plugin SDK
Pieces provides first-party plugins for:
- VS Code — Inline snippet saving, context recall, and AI-tag search directly in the editor.
- Chrome Extension — Captures links, highlights, and browsing context automatically.
LLM Configuration
# Example: configuring a local or cloud LLM provider in Pieces
provider: ollama # or openai, anthropic, gemini, etc.
model: llama3 # model slug
api_key: <YOUR_KEY> # optional, for cloud providers
on_device: true # default; set false to allow cloud routing
Key Capabilities
- LTM-2 Engine — Proprietary memory engine with up to 9 months of indexed, time-queryable context.
- Semantic & temporal search — Query memories by natural language, time range, app source, or AI-generated tags.
- Air-gapped by default — All processing happens on-device; cloud is opt-in per session.
- Bring-your-own-key — Supports OpenAI, Anthropic, Google, and local model providers.