explore
swipe
Please turn off low power mode for a smoother experience.
Artificial Intelligence

Developing faster without getting sloppier

We use AI in our own development work: for routine tasks, tests and migrations. Whatever comes out of it is reviewed by people before it reaches your project.

What this is about

AI does not change what makes good software, but it changes how much time the unpleasant parts cost. Tests, migrations and understanding someone else's code go considerably faster. The responsibility for the result stays with us.

Fast is not the goal

Generated code that nobody understands is more expensive than code written by hand. We only take on what we can be accountable for ourselves.

Strong with legacy code

Getting to the bottom of a system that has grown over years used to take weeks. This is exactly where using AI saves the most time.

Tests finally get written

Test coverage usually fails on effort. When that drops, tests appear where they would never have appeared before.

Code is written faster. Responsibility is not.

Situations that typically bring clients to us

What we actually do

Opening up legacy code

Analysis of applications grown over years, to make structure, dependencies and risks visible before anyone works on them.

Building tests

Automated tests for existing features, so later changes do not break something unnoticed.

Migrations and rebuilds

Recurring changes across many files, prepared automatically and reviewed by us.

Data preparation

Matching, cleaning and mapping large volumes of data when moving between systems.

Documentation

Technical documentation that actually gets written, because the effort no longer stands in its way.

The review stays human

Everything that goes into your system is read, tested and vouched for. Suggestions are suggestions.

How we work

  • 01

    Assess

    We check where in the project this is worth doing at all and where it brings nothing.

  • 02

    Secure

    Before anything is rebuilt, tests are written so changes can be verified.

  • 03

    Implement

    Routine work prepared automatically, technical decisions made by hand.

  • 04

    Review

    Code review and testing as in any other project, no exceptions.

The gain comes with the legwork, not with the architecture. Tell us which system is coming up at your end.

Let's connect 🚀

Frequently asked questions

Code is written faster, responsibility is not. We build custom AI applications when standard tools cannot carry the process.

What does dxm use AI for in development?

Above all where effort used to stand in the way: opening up legacy code grown over years, building tests for existing features, preparing recurring rebuilds across many files, matching large volumes of data for a migration and producing technical documentation. Getting to the bottom of a system nobody knows any more used to take weeks – that is where the time saved is greatest.

Does this produce code that nobody understands?

Not with us, because that is precisely what would be more expensive than code written by hand. Everything that goes into your system is read, tested and vouched for by us; suggestions remain suggestions. Speed is not the goal, it is a side effect. We only take on what we can still explain and develop further in two years' time.

How do you ensure quality?

Through automated tests that only come about because the effort is lower, through logging and spot checks in operation, and through clear error messages instead of a black box. With migrations and data matching we additionally check against real records, not just the ideal case. That way it stays traceable what the system does and where it reaches its limits.

Can this be connected to existing software?

Yes, and usually that is the decisive question. An AI feature that requires a second window is not used by anyone after two weeks. So we connect it through interfaces where the work already happens – in the CMS, the shop, the CRM or the inbox – and make sure permissions and data access follow the existing rules.

A system nobody knows any more?

Tell us what is coming up. We will tell you how quickly we can work our way into it.