Can one person build an AI MVP?

Yes — one experienced design engineer can ship a production AI MVP in 4–8 weeks. ProUX (12,000 users) and the AI Agent Sprint engagement on surinder.design are both built on this model: one person handling research, design, prompt architecture, and frontend code end to end, no handoffs.

The reason one person can ship an AI MVP faster than a four-person team is that AI-native products eliminate handoffs. There is no separate spec, design, build, and deploy phase — the deploy IS the design, iterated daily against real LLM output. A solo engineer-designer collapses six weeks of meeting and translation overhead into actual building.

What 'one person' realistically ships in four weeks: one production agent integrated with your stack, basic memory and retrieval, a clean chat or workflow UI, and the deployment pipeline. What it does not ship in four weeks: a multi-agent orchestrated system (that's the AI App Foundation tier, 8–12 weeks), heavy backend rewrites, or a full design system from scratch.

The risk of solo work is bus-factor — but it's mitigated when the deliverable is in your repo, on your infrastructure, with documentation. The deliverable on every surinder.design engagement is exactly that: working code in your repository, with a 30-day post-ship support window for handoff to your team.