
Velyr
AI agent that finds & fixes your conversion leaks.
Across my last few SaaS projects, I absolutely hated one thing: constantly optimizing my website. It swallowed massive amounts of time, and there was no software out there that could automate it, your only real choice was hiring an expensive agency.
I wanted a better way, so I built Velyr. It's a CRO agent that finds and fixes conversion leaks directly in your code. To keep things safe, you just approve changes with one tap via Telegram, and it auto-reverts instantly if performance drops (works with any GitHub-connected website and Shopify stores).
About
I hate digging through PostHog and don't trust AI in production. So I built a CRO agent that finds and fixes code leaks safely: you approve with one tap via Telegram, and it auto-reverts if performance drops.

3 Comments
What stood out to me is that you're shifting optimization from something people do to something they supervise.
That's a much bigger change than automating CRO. If users trust the approval and rollback process, the product becomes about reducing decision fatigue, not just improving conversion rates.
Exactly, that’s actually the direction I wanted to move towards.
The problem I experienced wasn’t that I didn’t know what to optimize, it was that CRO became an endless loop of analyzing data, creating hypotheses, implementing changes, and checking if they worked. It’s a lot of context switching.
The goal with Velyr is to turn that workflow from “doing optimization” into “supervising an optimization agent”. The approval step is there because giving an AI direct access to production code requires trust, and the rollback system is what makes experimentation safer.
I think the interesting question now is how much autonomy people are comfortable giving these agents over time. Thanks for the perspective!
Glad it resonated.
Your reply made me think there's one strategic decision sitting underneath that trust model which becomes much more significant as the product grows, but I don't think I can explain the reasoning properly in a thread without oversimplifying it.
If you're interested, what's the best email to reach you on?