Build software & cloud
I introduce development with AI agents (agentic coding) in your team, review architecture, cloud and code at a fixed price, and deliver products and platforms on your behalf.
of respondents in large organisations use AI coding tools.
have full governance in place for them.
are concerned that AI tools introduce new security vulnerabilities.
Source: Black Duck, June 2026 · 831 respondents in organisations with 500+ employees
Everyone has access to AI assistants, yet delivery time is the same as before.
Without shared rules, quality varies and reviews take longer instead of shorter.
Security, data protection or the works council want to know which rules govern AI-written code on its way to production.
This is how I build software myself: every task passes through the same stations. Whatever fails a gate does not move on.
Agents work in fixed roles: requirements, planning, design, code and verification. Every task passes through the same stations.
Tests, type checks and security scans run automatically. Whatever fails a gate does not move on.
No merge without review and approval. Accountability stays with your team.
The run is automated end to end and stops only when a gate is not passed. Approving the merge always stays with a person; plan approval and the smoke test are switched on depending on risk.
Reference flow from my own development. Stations, gates and thresholds follow your project; further steps are added where your project needs them.
Every role has its own rules, checkpoints and clear handovers. Legal flags topics; the legal assessment stays with a person.
Four examples. Further workflows are built from the same roles, tailored to your project and the way you work.
You start with an assessment and decide afterwards whether and how to continue.
We look at your situation: team size, tools, and what has changed since AI tools were introduced.
I review tools, process and pipeline and name the risks. You receive a target picture and a 90-day plan.
One team works with rule set, roles and gates on its own code. The approval process is proven in daily work.
The way of working moves on to further teams. The rules are maintained and the impact is measured.
An independent view with report, risks and roadmap. I do not sell cloud contracts or licences.
Cloud architecture assessment: each additional workload €2,400. All prices exclude VAT. The offer is aimed exclusively at businesses.
For projects with an open scope I work under a framework contract with agreed goals: billed by the hour, monthly statement, with a cap on request.
By effortBilled by the hour under a framework contract, with a cap on request. The rate is stated in the offer.
Tick what already applies today.
Tick what applies. Your result appears here.
Discuss the resultThe evaluation stays in your browser; nothing is transmitted.
That depends on your data and contracts. In the assessment we record which data may go where and configure the tools accordingly.
Both need clear rules and traceability. That is exactly what the rule set provides: who approves, what is logged, and what is not evaluated.
None in particular. I work with Claude Code and Cursor myself. The approach also works with the tools you have already licensed.
You do, along with everything created for you in the project. The rule set I bring stays mine; you may use it in your company permanently.
In addition to licences there are usage-based costs for the models. In the assessment we estimate them for your team.
In the intro call we look at your situation: team size, tools, architecture and what prompted the question. Afterwards you receive an offer within one working day, or a clear recommendation to leave it.