What Is Pieces and What Does It Do?

Pieces is an AI memory layer that runs in the background on your computer and builds a searchable record of your work as you go. It captures activity from the apps you already use — email, chat, browsers, docs, design tools, and AI assistants — so you can recover decisions, resume where you left off, and bring real context into other AI tools. It's most relevant if your work is spread across many apps and you regularly lose time reconstructing what happened or what you were about to do.

The core idea: memory instead of manual note-taking

Most productivity tools ask you to record things deliberately. Pieces works the other way around: it forms memories automatically from what you're already doing, then makes that record searchable and reusable.

According to the site, Pieces forms memories every 2 seconds from the apps you already use. That cadence is the point — the memory is meant to keep up with a normal working day rather than depend on you stopping to document it.

The stated purpose is to turn "what already happened" into "what you need to understand, do, or create next."

What it captures

Pieces organizes capture into a few categories, and the distinction matters because not everything is collected the same way:

The copy-and-paste channel being opt-in is worth noting if you handle sensitive text — that stream isn't on unless you turn it on.

The source list is broad and spans both work and personal tooling. Examples named on the site include:

• Email and messaging: Gmail, Outlook, Slack, Teams, Google Chat, Discord
• Browsers: Chrome, Arc, Safari
• Meetings: Google Meet, Zoom, Microsoft Teams
• Docs and planning: Google Docs, Notion, Figma
• AI tools: Claude, ChatGPT, Copilot, Codex, Perplexity
• Development: Cursor, VS Code, JetBrains, Xcode

For each, the site describes the kind of context captured — for example, Gmail yields threads, follow-ups, decisions, and account context; Slack yields team threads, handoffs, decisions, and open questions; Figma yields design reviews, comments, rationale, and assets.

What you can actually do with it

Three capabilities follow from having that memory:

Search and revisit any moment. The captured activity forms a chronological timeline you can search, filter, and revisit. You can ask questions about what you've done and scope the query by time, source, or topic.

Generate summaries from real context. Standup updates, morning briefs, meeting prep, and day recaps can be produced from your actual activity rather than from memory or guesswork. The site frames this as turning "hours of retracing your steps into one click."

Carry context across tools. Because the memory sits underneath your other applications, context can follow you between them — including into AI tools that would otherwise start from nothing.

Privacy and security

The site lists privacy and security as fundamentals rather than features. The concrete detail available is the capture model described above: focused-application capture is on by default, copy-and-paste capture is opt-in, and meeting capture comes from conversation apps and sources. If your work involves regulated or confidential material, that capture model is the first thing to evaluate against your own requirements.

Where it fits — and where it may not

Pieces is aimed at people whose work generates a lot of scattered context: researchers, engineers, designers, account and support teams, anyone living in meetings plus chat plus docs plus an AI assistant.

It's less obviously useful if your work already lives in one system with good native history, or if you can't have background capture running on your machine for policy reasons.

Two things the site references but this overview doesn't resolve: Pieces MCP and teams & enterprise offerings. If you need to connect Pieces to an MCP-compatible client or roll it out across an organization, those are the pages to check next, along with pricing at pieces.app/pricing.