We read the system before recommending one
We look at the actual configuration, pipelines, and logs, not only the documentation and interviews. The current state is usually different from the diagram.
About CogniLevel
The people who draw the architecture also build it.
CogniLevel is an engineering consultancy for AI, data, and cloud systems, based in Belgium. We don’t hand over a slide deck and leave. We stay to write the code, run the first restore, and make sure your team can carry on without us.
Why we work this way
Most architecture problems show up during delivery.
A design can look clean on a whiteboard and still fall apart on contact with a real system: a source table with no stable key, a firewall rule nobody owns, or an AI assistant that can suddenly read HR files. When the people who made the design are also the ones building it, those problems come back to them rather than to whoever inherited the plan.
That is also why we cover AI, data, and cloud together. An AI workflow depends on the data it reads, and the data depends on the platform underneath. Treating them as three separate projects with three suppliers is how gaps between them go unnoticed.
We work mostly with organizations in Belgium and take on projects elsewhere in Europe. Engagements range from a two-week architecture review to building alongside your team for several months.
What we hold ourselves to
Three habits we don’t skip.
We write decisions down
Every significant choice goes into an architecture decision record: the options, what we picked, and why. When circumstances change, your team can see which assumption to revisit.
We test the handover
Before we leave, your engineers deploy, roll back, and restore the system themselves. If they can’t yet, we aren’t finished.
Contact
Tell us what you’re working on.
A few lines about the situation is enough. We reply with the questions we would ask first, and an honest view on whether we can help.