Pieces
On-device AI memory layer for developers — captures context, manages snippets, recalls anything.
About
Pieces is an on-device AI assistant for developers that acts as a memory layer across your workflow. Its background service captures what you see and copy across the IDE, browser, and chat tools, turning it into a searchable timeline — ask by time, person, or topic and pull a recap in one click. It also does AI-enriched snippet management: save code, and Pieces adds titles, descriptions, and metadata automatically.
Privacy is the pitch: processing and storage happen on-device by default, air-gapped from the cloud unless you opt in. An MCP server carries your captured context into Claude, Cursor, Codex, and other MCP clients. Founded in 2020, US-based.
Worth knowing: the genuinely useful tier is Pro at about $18.99/month — the free plan's memory is a rolling time-bound window, and long-term memory (up to 9 months) is a Pro feature. Background screen capture is a real trust decision even when it stays local, so review the pause/block/delete controls before enabling it broadly.
What it can do
- On-device capture of code, browser, and chat context
- AI-enriched snippet save, search, and transform
- Time-based memory search with natural-language queries
- MCP server for Claude, Cursor, and Codex
- Local LLM support with optional cloud models
Setup
Download PiecesOS for macOS, Windows, or Linux from pieces.app. Install the IDE and browser plugins for capture.
Prerequisites
Desktop OS (macOS, Windows, Linux); optional cloud account for sync.
Use cases
Works with
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 45; Trending score 68/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.