Clarus.dev's answer:

Every capture is indexed by the text inside it. Text recognition runs locally the moment you take a screenshot or finish a recording, so months later you can search for a word you remember seeing and get the picture straight back. Most capture tools leave you with a folder of files named by timestamp; Clarus leaves you with something you can actually search.

It also takes over Win+Shift+S, so it lives where your hand already goes rather than asking you to learn a new shortcut.

Clarus.dev's answer:

Most Windows users are really choosing against the Snipping Tool, which they already have. Clarus does everything it does, then keeps the capture useful: a searchable history, screen recording on a real editing timeline, annotation for both screenshots and video, and share links you can expire or password-protect. The Snipping Tool captures and forgets.

Against a paid tool like Snagit, the differences are that text recognition runs entirely on your machine rather than in a cloud, the free tier is permanent and needs no account, and nothing leaves your computer unless you explicitly choose to share it.

Clarus.dev's answer:

People on Windows who take a lot of screenshots as part of their work and keep losing them: developers filing bug reports, support and QA teams, designers collecting references, writers and teachers making documentation, and anyone who screenshots a receipt, a booking confirmation or an error message and needs it again weeks later.

It is built for individuals and small teams rather than large IT-managed rollouts.

Clarus.dev's answer:

It started as a personal annoyance. I took screenshots constantly and never found them again — a folder of thousands of files named by timestamp is not a record of anything. The obvious fix was to make the pictures searchable by what is written in them, which meant running text recognition locally on every capture instead of shipping them off to a server.

Everything else grew out of using it daily: the annotation editor, then screen recording, then a real video timeline, because trimming a recording in something else was slower than building the tool. Clarus is made by one person in the UAE.

Clarus.dev's answer:

Rust and Tauri 2 for the native Windows layer, React 19 with TypeScript for the interface, and Vite for the build.

Capture uses Windows Graphics Capture and DXGI Desktop Duplication; recording encodes through Media Foundation with hardware H.264 where the GPU supports it. Text recognition is PaddleOCR running locally through tract, a pure-Rust inference engine, with the Windows OCR API as a fast first pass.

The cloud side is Supabase with Cloudflare Pages.