vipin-tanna

Short demo (browser + voice agents): https://youtu.be/BA0KU69Yv_w

Happy to share internals and take bug reports in the thread.

Happy to share technical guts: we capture actions, bind to resilient DOM anchors, expose the flow as an editable step list, and support parameterized inputs + loops. Roadmap: better selector heuristics, headless/cloud runners, approvals, and versioned flows. Known gaps: CAPTCHAs/2FA, anti-bot walls, highly dynamic UIs. Send me your worst sites.

We call it Gabriel Operator—record a task once, get an agent you can reuse/share. It operates in your browser, not via vendor APIs. We built this after getting tired of scripting the same “download this report, paste that ID, submit those 12 fields” flows across dozens of portals.

What it’s good at: repetitive CRUD, back-office portals, weekly reports, job/lead sourcing, invoice/status checks. What it’s bad at: CAPTCHAs, 2FA, wildly changing DOMs, anti-automation defenses.

If you try it, I’d love bug reports and “this broke on Site X” details. We’ll be in the thread all day.

Link: https://gabrieloperator.com

We just launched Gabriel Operator — a browser-based platform that transforms your screen activity into fully executable AI agents. You simply record what you do in the browser, and we turn it into a repeatable agent that runs like a workflow—no APIs, no code.

What makes it different? • Works entirely off your screen (great for websites without APIs) • Agents can loop, branch, and request input • You can publish & monetize your automations • Fully web-based, no installation needed

Next week, we’re rolling out Creator Mode — allowing anyone to publish, share, and sell their own “Digital Angels” (our name for agents). We’re offering free access for the first month to early adopters who want to shape how this platform grows.

Would love feedback, thoughts, or AMA. We’re building this for real-world automation beyond devs — and we’re still early.