FusedFrames documents your work while you do it. The desktop app records what you do on screen, and FusedFrames splits it into one recording per task, a timeline of readable steps. Document libraries write the recordings you choose into documents that follow your template, and questions asked along the way capture the reasoning behind the steps. Your team reads those documents in the dashboard and the desktop app, and your AI agents read them over MCP, the CLI and the API.
Written documentation contains what someone remembered to write down, and it goes stale the moment the work changes. Documentation made from the work itself can do more.
Written automatically
Documents are produced from what you actually do on screen. Choose Start recording, work as you normally would and choose Stop recording; nobody sits down to write. A document library writes the recordings that match its sources into documents that follow the template you set, automatically or after your review.
Reasoning included
Documents capture why you made each decision, not just the steps you took. When a recording holds a decision the steps alone cannot explain, a question card appears over your screen while you work. Type the answer or hold the talk button and speak it, or answer later on the recording's page. Answers are saved with the recording and written into the document.
Never goes stale
Documentation stays up to date as the way you work changes. When a new recording matches a task that already has a document, the document is updated rather than written again. Repeat the work and the write-up follows, so nobody has to notice that a page has gone out of date.
Ready for your AI agents
Everything FusedFrames writes is readable by your agents. The hosted MCP server needs nothing installed, the CLI is one small program installed from npm and the API is read-only REST. All three read the same documents, and the source recordings behind them when you allow it, with the same API keys, so they drop straight into Claude Code, Cursor or any script you already run.
Knowledge from everyone
Know-how comes from the whole team, not trapped in one person's head. A recording stays private to the device that made it until someone shares it to the workspace. Shared recordings from every device a library's saved filters cover feed the same documents, so people doing the same task update one write-up between them. Each document says how many times its task has been seen and on how many devices, so you can tell settled habits from one-offs.
Rare cases included
The odd cases nobody thinks to write down are captured too. A written guide covers the common path, but the awkward variant someone handled once is exactly what the next person needs. If it happened while recording was on, it is in a recording, ready to be written into the document.
When you are ready, start here.