great project, Fazle. there are definitely a lot of agent browser tools, but none of them are solving the core regression testing problems. This is def going deeper into this domain.

Awesome work getting to 95 on mobile! The final 5% is almost always mobile JS execution time. A few quick wins: check your bundle with vite-bundle-visualizer, defer all third-party scripts, and test on real low-end devices. You’re super close! 🚀

Really interesting problem to solve—secure browser automation is a big gap right now. The relay approach and built-in token optimization sound promising. Curious to see how it performs in real-world use. Looking forward to the launch!

Great breakdown. Totally agree that code generation has outpaced testing. The “works in demo, breaks in prod” problem is very real, especially with auth and edge cases. Feels like the gap now is less about building and more about validating fast enough. We’ve started keeping a small set of critical flows tracked in Tuskr just to make sure nothing essential breaks as we iterate quickly. Not perfect, but it helps bring some structure back.

Welcome, Valentin Monteiro. It's nice to have you here. Hope to learn from you.

For me, an AI app feels genuinely useful when it saves time and simplifies tasks without making me think about how it works. Anything that feels like a gimmick usually overpromises or interrupts my workflow rather than helping.