Callaars & De Gruyter · Software development

Software that keeps running, after the demo too.

With AI you can have something working on your laptop in an afternoon. That's real progress, and we use those tools every day ourselves. But the work that makes software expensive starts after that: who may see what, what happens when it fails, where the data lives, and who maintains it a year from now. That's what we're good at. We've been building software since 2004.

Small or large, always into production.

MVP sprint. From idea to working product. The smallest version that actually helps a real user, hosted and usable.
Custom project. A larger project with a defined end point: a complete platform, not a single component.
Internal → external tool. Your concept already runs internally. We get it ready for customers, without budget surprises.
Dedicated capacity. An engineer who works inside your team, your tools and your backlog. No fixed scope.

Decide first, build after that.

Every project goes through the same four steps. After each one you have something in hand, even if you stop there.

Listen. A conversation with the people doing the work. What should it solve, and for whom? You get: An honest view: is this worth building, and are we the right fit.
Spec. We pin down the decisions you can't change later without rebuilding. You get: A document with the design, who may see what, where the data goes, the integrations, and a fixed price for the build.
Build. We build what's in the spec. Fixed price, fixed scope. You get: Working software in production, with monitoring, error handling and a handover to your team.
Care. Software grows. Every major new feature gets its own, smaller spec. You get: Support or dedicated capacity, with one fixed point of contact.

Fast is good. Fast and then again is not.

Vibe coding means building software by letting an AI tool write whatever you describe, without anyone really designing the code. For a prototype that's fine. For something customers or staff lean on every day, it runs into questions nobody asked up front.

fig. 01 · two roads to production

Access. Quickly assembled: Everyone sees everything, because that worked in the demo. Our approach: Defined per role: who may see and do what.
Data. Quickly assembled: Lives wherever the tool put it. Our approach: Decided up front where it lives and who signs for it.
Failures. Quickly assembled: You notice when a customer calls. Our approach: Logging and monitoring from day one.
Cost. Quickly assembled: Goes up without anyone noticing. Our approach: A ceiling and an alarm point in the design.
A year from now. Quickly assembled: Nobody remembers why it works this way. Our approach: The spec explains why, for whoever maintains it next.

Want to build it yourself with AI? You can. Then we write the spec, and check regularly that what you build stays within what was agreed. How that works for AI projects →

Tell us what you want to build.

A 30-minute call. You tell us what you have in mind, we tell you honestly how we'd approach it and whether we're the right fit.

Book a 30-min call · [email protected]