Prototype it anywhere. Build it here.
Your prototype from Lovable, Claude or a whiteboard becomes the spec. In six weeks we turn it into a secure MVP on twenty years of architecture. In three months it runs as a SaaS product in Europe, with a security report for your customers.
AI made the demo cheap. Your customer now asks for evidence.
Anyone can have a convincing app on screen by Friday. Then the first customer sends a security questionnaire, a tender asks for your ISO process and an investor asks who is accountable. That is where most products stall. It is exactly the stretch we have done for twenty years.
of security tests on AI-generated code, when the model was given no security guidance.
Veracode · GenAI Code Security Report · 2026The Dutch Cyberbeveiligingswet (NIS2) took effect. Covered organisations must check their software suppliers, so the questionnaires flow down the chain to you.
Cyberbeveiligingswet · in force 15 August 2026One week to scope. Six weeks to launch. After every step, on only when everything is green.
Coding agents do most of the typing, under Harness Engineering: the writer never reviews its own work, every change is tested and scanned, and a failed check stops the line. People decide what gets built and sign off what ships.
- BEFORE WEEK 1Sprint Scope, clickable prototype, architecture, fixed quote.Scope frozen
- WEEK 1Foundation Entry set up, cloud environments, automatic checks switched on.Live in test
- WEEKS 2–4Build Agents build your features on our blocks, with tests written alongside.Features done
- WEEK 5Harden Security scans, payments, a penetration test, OWASP ASVS review.Security sign-off
- WEEK 6Launch EU production, monitoring, handover to Run.You accept
From MVP to operational SaaS. First customers onboarded, billing running, monitoring and support in place, the next features through the same gates. At three months it is a product you sell, not a project you finish.
Six things AI tools alone don’t give you.
A foundation that already runs products.
Entry is our open-source .NET and Angular blueprint, and the base under our own products today. The agents inherit twenty years of architecture instead of inventing one per prompt.
Pre-hardened SaaS blocks.
Login, separate data per customer, payments, activity log, privacy features. Built once, tested once and reused, not regenerated from scratch with fresh bugs.
Agents that follow our rules.
Separate agents for spec, build, test, review and security, each working to Enigmatry guidelines, in isolated sandboxes in our Azure environment in West Europe, not on a laptop and not at a third party. The one that writes the code never approves it.
Code moves on only when every check is green.
Every change is built, tested and scanned for security holes. A separate job rebuilds it from scratch: the new tests must fail before the change and pass after it. A failed check stops it. Nobody merges on a feeling, and neither does an agent.
An architect signs off every release.
Against OWASP ASVS, inside our ISO 27001-certified process. That signature is what your customer’s security officer wants to see.
Ownership, and someone to call.
You own your product code and your data. We run it in Europe, with the source in escrow (EnigmaSure). When something breaks at seven in the morning, a person picks up.
Keep the prototype tool. Change who builds the product.
Lovable and Claude are brilliant for finding out what you want. We use coding agents ourselves: they are our engine, not our rival. What they do not give you is someone accountable for the result.
| Option | Build it yourself with AI | MVP agency | Dutch studio | SaaS Factory |
|---|---|---|---|---|
| Time to production | Days to a demo; production is up to you | Weeks | 10–14 weeks | 6 weeks MVP · 3 months operational |
| Login, tenants, payments, audit log | Generated per app | Usually built per project | Built per project | Pre-hardened blocks, reused |
| Security evidence | Platform scans, if any; you answer the questionnaire | Rarely published | Varies | Signed report, ISO 27001 process, OWASP ASVS |
| Who is accountable | You | Until delivery | The studio | We are, and we run it |
Four ways in. One proposal, made for your product.
No two products need the same build, so we do not print a price list. The Factory Sprint ends in a proposal with a fixed price for your scope, so you know exactly what you get before you pay for it.
Factory Sprint
- Scope, prototype, architecture
- Ends in a proposal with a fixed price
- Credited in full to your MVP
Partner MVP
- A monthly platform fee
- A share of what each end-customer pays you
- Platform licensed; buy-out after three years
Classic MVP
- You own all the code
- No revenue share
- Run contract after launch, monthly
Hardening Audit
- Review of an AI-built app
- Security and architecture findings
- Fix it, or rebuild it here
- Credited in full to a Sprint or rebuild
Scope changes: swap a feature for one of equal size at no cost. Anything extra is quoted separately or done from Run hours.
We earn when your product earns.
If you already sell into a niche, you know the customers, the language and the price. You should not have to fund the whole build before the first invoice goes out. With Partner you pay less up front, and we take a share of every end-customer you sign.
That keeps us honest after launch: our income depends on your product staying up, staying secure and selling. You own your product code and data. Our platform and blocks are licensed while usage runs, the source sits in escrow (EnigmaSure), and after three years you can buy the platform out.
- LOWER FEE UP FRONT
- MONTHLY PLATFORM FEE
- A SHARE PER END-CUSTOMER
- BUY-OUT AFTER 3 YEARS
Three kinds of people walk in. All three leave with a product.
The Lovable graduate.
Your AI-built prototype works and people want it. Then a customer sends you a security questionnaire. Start with a Hardening Audit, or go straight to a Sprint.
The domain expert.
You are a Dutch or Belgian company of 20 to 500 people. You know a niche better than any software house, and you already sell into it. Software is not your core. Partner was made for you.
The existing client.
You know how we work. You have an idea that should be a product instead of another module. Same people, same gates, a different shape of contract.
- A product owner with at least four hours a week who can decide.
- A budget for a six-week build, or a sales channel that makes Partner worth it.
- A trigger: a security questionnaire, a tender, an investor check, a NIS2 supplier review, or an outsourced build that failed.
- A package already does it. We will name it and send you there.
- You want a prototype, not a product. Lovable is genuinely good at that, so use it.
- Nobody on your side can make decisions during those six weeks. Without that, it will not be six weeks.
The questions everyone asks.
Why not just keep building it with Lovable or Claude?
Do, for as long as you are finding out what to build. The moment real customers, real data and someone else’s security officer are involved, the question changes from “does it work” to “who answers for it”. We have an answer to that: the same agents work inside gates, on a foundation that already carries products, with an architect who signs off.
Do you reuse my prototype’s code?
We reuse what it taught you: the screens, the flows, the words your users understood. The product itself is rebuilt on Entry and our blocks, because that is where the security and the speed come from. Your prototype becomes the clearest spec we have ever been handed.
Six weeks, really?
For a scoped MVP on our blocks, yes, and we are proving it on our own product first, with the timed log published. A product that needs more than six weeks of features gets them after launch, through Run: live, with customers, instead of in a longer project.
Who owns what?
Classic: all the code is yours. Partner: your product code and all your data are yours. Our platform and blocks are licensed while usage runs, with the source in escrow and a buy-out after three years. Either way, Entry itself stays open source under Apache-2.0.
Why .NET and Angular?
Because they are boring in the best way: long-supported, widely known, and readable by the developers you may hire later. Your product should outlive the tool that sketched it.
Have a prototype? Bring it.
Show us what you built with Lovable, Claude or on paper. In one conversation you hear whether it fits the factory, which offer suits you, and what we would leave out. The whole argument — why a prototype is a spec, not a product — is in the launch article.
Factory Sprint · five days · credited to the MVP