An AI product built by an AI-run delivery process

Pebblearning is AI on both sides of the work. The product itself uses AI to take the routine manual work off specialists in learning diagnostics, the kind of repetitive processing that eats their hours and adds nothing to their judgment. And the product is built the same way, by an AI-run delivery process where I’m the co-founder carrying the whole technical side as the only human in it.

Every feature starts as a spec that agents can act on, then moves through agent-written code, automated verification and deployment, with my review focused on intent and product behavior rather than on syntax. It is the same methodology I set up for a full R&D org, compressed to a single person.

On the money side I kept the early-stage bill where it belongs, near zero. Cloud startup credits carry the infrastructure, and the stack is sized to scale with real users instead of paying for capacity on faith. Pebble is equity work too, I joined as a co-founder rather than a contractor, which is exactly the model behind my equity partner track. What the automation does for the specialists, and how the one-person flow holds up as the product grows, I walk through on a call.