Lovable reached $200M ARR faster than any European startup in history, closing a $330M Series B at a $6.6B valuation. Bolt.new hit $40M ARR in six months. v0 by Vercel has become the default for frontend UI work in the Vercel ecosystem. These aren’t niche experiments — they’re genuine businesses with enterprise customers including Klarna, Uber, and Zendesk.
The practical consequence isn’t just that more people are prototyping with AI tools. It’s that founders and product leads are arriving at commissioning conversations with something they didn’t have two years ago: a working reference prototype.
How the briefing has changed
Before tools like Lovable and Bolt, the typical brief for a custom app was a mix of wireframes, written specs, competitor examples, and verbal description. The gap between what was described and what was meant was significant, and a lot of early project time went on bridging it.
With a Lovable or Bolt prototype in hand, that gap closes considerably. Founders can point to a working demo that shows user flow, core interactions, and basic information architecture. The brief becomes a conversation about what needs to change, what the production version needs to handle that the prototype doesn’t, and what the prototype got right that’s worth preserving as a design reference.
What prototypes are good for, and what they’re not
The tools work well for: proving a concept fast, validating user flows with real feedback, investor demos, and narrowing down scope before a professional build.
They hit limits at: proper authentication and session management, third-party integrations that go beyond basic API calls, GDPR-compliant data handling, NHS or regulated-sector integrations, performance at scale, and anything where the codebase will be maintained and extended by a team over time.
The pattern we see most often is: AI prototype to validate, professional build to ship. The prototype is the map; the production codebase is the territory. The map is useful — it just isn’t the territory.
For projects that have outgrown what the prototype can do, our custom software development and AI-assisted development pages cover how we work.