vfrog

Three years ago, I switched from corporate finance to self-taught CS / AI engineering. This led to Planow.ai — a retail shelf analytics founded 2 years ago with 2 experienced hackers. We had real clients (Arla, Red Bull), 95% accuracy, but couldn't scale. We were ready to quit.

Then a competitor asked to license our models. That's when I realized: the value wasn't the models. It was the infrastructure.

Only 1% of developers can build and deploy computer vision. But cameras are everywhere now—phones, smartglasses, robots, IoT. The bottleneck is clear: we need better infrastructure.

So we pivoted to vfrog. Infrastructure for computer vision.

Our team: Fred (CTO, 30+ years Infrastructure), Jonas (CAIO, 25+ years AI/ML/CV). Not junior engineers.

Traction: 3 enterprise clients, $2k MRR, 5 months in, bootstrapped.

Now we're launching self-serve: 2-week free trial, 500 credits, no commitment. We also built an npm package for Claude Code so you can deploy CV inside your AI workflow, completely free.

Quick question for you: Did you ever try to build CV into your application? What stopped you?

And if we made it frictionless, what use case would be most valuable? Real-time detection? Document understanding? Custom classification?

Drop your thoughts below. We're listening.