AI prototype to production
You used Lovable, Bolt, Replit, v0 or Cursor to build something that works. The gap between a demo and something you can put real customer data into is work the demo never shows you: who can see which records, where the keys live, and what happens when something breaks. I build that layer in afterwards, step by step, while the application keeps doing what it already did.
Based in Groningen, working across the Netherlands and remotely for clients worldwide.
Turning a prototype built in Lovable, Bolt, Replit, v0 or Cursor into software you can safely put real users and real customer data on.
Suitable when something already exists and works, but nobody can say who can see which records or what the next change will break. In practice that usually means any logged-in user can read every other customer's data.
Good for
- Apps from Lovable, Bolt, Replit or v0 that have to go live with real customer data, or code you wrote yourself in Cursor.
- Prototypes where nobody can say with certainty who can see which records.
- Applications where every change breaks something elsewhere, and you spend more time fixing than adding.
- Software that was built with AI and now has to comply with the GDPR, including where the customer data physically sits.
How I approach it
- A review first. I get the application running, read through it, and set out what's sound, where the risks are, and the order I'd work in. I need access to the code and to wherever it runs, nothing more. Once I have that, the report follows within about a week. Scoped work, usually a day or two, at a fixed price. The report is yours even if you carry on by yourself afterwards, or with someone else.
- Sharp edges come first. Keys that ended up in the browser get replaced, and the new ones move to the server. The rule about who can see which records moves to where a user can't touch it. And the backup gets restored once, to check that it actually works. If you already have users on it, this comes before everything else.
- Then the foundation. You get database changes you're still willing to make once there's real data in there, error handling in the places where things genuinely go wrong, an alert the moment something breaks, and tests around the behavior that must not change. Each step is a milestone with a fixed price, and you can stop after any of them.
- And then it's yours. The code lives on your GitHub, runs on your hosting and uses your accounts, including halfway through. What I learn along the way I write down there, so you don't end up depending on one person who happens to know how it works. Me included.
Tech I work with
Supabase, Postgres, row level security, edge functions, TypeScript, React, database migrations, characterization tests, GitHub Actions, error monitoring, backup and restore, Cloudflare. The apps themselves usually come out of Lovable, Bolt, Replit, v0 or Cursor.
Frequently asked questions
Where will my customer data be stored?
Wherever you need it, and inside the EU if your customers or your regulator require that. This matters more than it sounds: on most of these platforms the region is set once when the project is created, and changing it later usually means a new project and a migration with downtime. I work from the Netherlands, so EU hosting is the default rather than an exception.
What does it cost to take an AI prototype to production?
It depends on what's there, so I won't name a figure before I've seen the application run. The review is scoped work, usually a day or two, at a fixed price you know in advance. What's needed after that goes in the report, with a fixed price per milestone, and most milestones are one to two weeks of work. Sending a message costs you nothing, and neither does my first reply.
Will anyone else see my code?
No. I don't name clients and I don't show code, yours included. That's also why there are no examples of rescued prototypes on this page. If you want it in writing, I'll sign an NDA before I look at anything. And you don't need to tidy up first: I'm looking at what's there, not at how it looks.
What do I get after the review, and is it mine?
A report covering what's sound, where the risks are, what I'd do, in what order, and roughly what that costs. The report is yours. If you'd rather carry on by yourself or with someone else, it's enough to hand to whoever does the work.
Do I have to commit to the whole thing at once?
No. After the review we work in milestones with a fixed price each, and you can stop after any milestone. The code lives on your own GitHub and is yours, including if we stop halfway.
Do I have to rebuild it from scratch?
Usually not. The valuable part of your application is the decisions built into it: which screens exist, what users do and which rules apply. That part stays. If the review shows that rebuilding really is the better option, I'll say so, but only after I've seen the code and not on a guess.
Does this work if I didn't build it with Lovable or Bolt?
Yes. Lovable, Bolt, Replit, v0 and Cursor come up most often, but what matters is what was built, not what it was built with. If you worked in Cursor on your own machine, the code is already yours and some of the work falls away.
Send me what you have
A few sentences about what you built and where it hurts is enough. I'll reply within a working day with what I think, and it costs you nothing. No need to tidy up first.
Get in touchOther services
- Custom application development End-to-end design and implementation of bespoke applications across web, native iOS, and backend systems, including MVP builds for early-stage products.
- Web, API and backend development Production websites, web apps, and APIs in React, Preact, Svelte, and vanilla web. Backends in Go and Python with REST or event-driven APIs.
- AI and LLM integration LLM-powered features for existing apps and new products: agent workflows, RAG pipelines, and integrations with OpenAI, Anthropic, and similar providers.
- Legacy code modernization Refactoring and modernizing older codebases. Framework upgrades, careful test-driven changes, and migrations from end-of-life platforms.
- Native iOS apps Native iPhone and iPad apps in Swift and SwiftUI, including widgets, Live Activities, and App Store delivery.
- Browser extensions and integrations Chrome extensions, Thunderbird add-ons, and integrations between browsers and external services.
- Point-of-sale and retail software Custom POS systems, inventory and stock management tooling, and integrations with receipt printers and other store hardware.
- Performance-critical engineering Algorithms, audio and video processing, codec work, P2P networking, and other engineering-depth projects where performance and correctness matter.