Custom software & AI development

AI Engineering

Merlin designs, builds and ships AI solutions that run inside your organisation. From the first working version to a product your teams use every day.

Merlin Engineering

Three ways to start

Start where you are. Each stage stands on its own, and each one leads to the next.

2–4 weeks

Prototype

Test an idea on real data before you invest in it.

A working version built on your data

Tested with the people who will use it

A clear go or no-go, with the next steps

1–2 weeks

From localhost to production

Already built something with Claude Code or Cursor? We take it to production.

Code and security review

Authentication, hosting and monitoring

Deployed in your cloud, with documentation

First release in 6–8 weeks

Product Studio

Turn a validated idea into a product your teams rely on.

A first release running in your cloud

New features every two weeks, shown in demo sessions

Documentation and handover built in

The Merlin Software Factory

An agentic framework our engineers built to deliver faster without lowering the bar.

AI does the heavy lifting. People own the decisions.

A product roadmap next to an AI interview that challenges a feature before any code is written

Client story • hub.brussels

From a pile of documents to a daily signal

hub.brussels wanted to know which international companies are setting up in Brussels. The answers were there, buried in thousands of filings in the Belgian National Gazette. We built a pipeline that extracts and structures them into a clear, daily view of foreign direct investment in the region. Automation and AI where they add value, so the time went where it matters: testing, security and validation.

3 weeks

to build, from first conversation to a running pipeline.

~22,000

documents processed every month, continuously.

Less than a coffee a day

to run, in production.

What stays with you

Your code, from day one.

Everything lives in your repository. No lock-in.

Your cloud.

It runs in your own cloud environment, under your control.

Your documentation.

Every feature is documented as it ships, written for people who don’t code.

Support, if you want it.

We can keep running and improving it, or hand it over to your team.

From idea to adoption

One team, end to end.

FAQ

Frequently asked questions

How long does it take to build an AI prototype?

Two to four weeks. You get a working version built on your own data, tested with the people who will use it, and a clear go or no-go with the next steps. It lets you test an idea on real data before you invest in it.

We built an app with Claude Code or Cursor. Can you take it to production?

Yes, usually in one to two weeks. We review the code and its security, add authentication, hosting and monitoring, and deploy it in your cloud with documentation.

What is the Product Studio?

The Product Studio turns a validated idea into a product your teams rely on. A first release runs in your cloud within six to eight weeks, new features follow every two weeks and are shown in demo sessions, and documentation and handover are built in.

Who owns the code, and where does the solution run?

You own the code from day one. Everything lives in your repository, with no lock-in, and the solution runs in your own cloud environment under your control. After delivery, we can keep running and improving it, or hand it over to your team.

How does Merlin use AI to build software, and who checks the result?

We use the Merlin Software Factory, an agentic framework built by our engineers. Every feature is challenged before any code is written. AI agents then write the code and its tests within our engineering standards. Automated checks and AI reviewers test every change, and an engineer reviews it. Nothing ships without a human sign-off.

Do you have an example of an AI solution running in production?

For hub.brussels, we built a pipeline that extracts and structures filings from the Belgian National Gazette into a daily view of foreign direct investment in Brussels. It took three weeks from the first conversation to a running pipeline, processes about 22,000 documents a month, and costs less than a coffee a day to run.

Have something worth building?

Tell us what you want to build, or what you’ve already built. An engineer will get back to you.