How We Built 30 AI Products on One Platform
A deep dive into the engine registry architecture that lets us ship a new vertical every week — from search to finance to code agents — all sharing one data layer and auth system.
The thesis behind neww.ai was simple: one AI platform, thirty vertical products.
The technical bet was that a well-designed engine registry could let a small team ship as fast as an army. Eighteen months in, the pattern holds. Every product — Search, Shop, Finance, Code, Health — shares one auth layer, one billing system, one data fabric, and one agent runtime. When we ship a new engine, every product gets it for free.
This post walks through the architecture: how the registry works, how we resolve capabilities across products, and the trade-offs we accepted to move this fast. If you're building anything with more than one AI use-case in it, this should save you months.