AI-native software engineering

25 years of engineering experience. Delivered at agent speed.

I'm Bert Groenveld, software architect. I design and ship production software with AI agents as my team, and with the engineering discipline that proves it works.

  1. Humanspec
  2. AIagents build
  3. Automatedtests
  4. Automatedquality gate
  5. Humanreview
  6. Automatedship

How a change reaches production: I specify and review, agents build, the pipeline verifies.

Services

Three ways to work with me.

01

Build

Production software, built with AI agents.

Greenfield products and new features on .NET and Azure, delivered by one senior architect directing a team of agents. Short feedback loops and working software early.

02

Review

An independent check on AI-written code.

Is the code your team generates with AI any good? I review it for quality and maintainability, and set up the quality gates in your pipeline that keep the standard up from then on.

03

Enable

An AI-native team.

Workflows, guardrails and review practices that let your developers work with agents without lowering the bar. Hands-on, in your codebase and on your pipeline.

Approach

AI writes the code. The pipeline proves it.

Generating code is cheap now. Knowing whether it is right is not. These are the guardrails I have relied on for years, and they are exactly what makes agent-written code trustworthy.

  1. Specs before prompts

    Behaviour is written down first, as executable specifications and end-to-end tests. Agents build against them, so "done" is something a machine can check.

  2. Quality gates, not opinions

    Static analysis with SonarQube or SonarCloud runs in the build and release pipeline as a hard gate. Code that misses the standard does not ship, whoever or whatever wrote it.

  3. Everything is code

    Software, infrastructure, documentation, tests and database schema all live in Git. Every agent change is a reviewable diff, and you can always go back.

  4. A human signs off

    I read what ships and I am accountable for it. Robust, testable and maintainable, pragmatic where it helps, and never losing sight of business value or the delivery date.

About

Portrait of Bert Groenveld
Bert Groenveld Microsoft Certified Professional Certified Scrum Master LinkedIn

Architect first. Early adopter always.

Developer since 2001, architect since 2016. I started as a software engineer at a product supplier, grew into lead developer, and from there into software designer and solution architect. More than twenty years on the Microsoft stack.

I love to create something new. Greenfield projects, so to speak. And I have always been an early adopter, curious how new technology works and how it can help my client. Working AI-native is the logical next step.

Since 2025, AI agents have been a fixed part of how I deliver. I direct agents such as Claude Code and OpenAI Codex, and spend my own time on specifications, architecture and review.

What has not changed: I care about code quality, I would rather discuss possibilities than move around obstacles, and I deliver on time.

Platform
.NET Azure Azure DevOps Blazor WebAssembly Angular
Architecture
Ports and adapters CQRS Messaging End-to-end testing SonarQube / SonarCloud
AI agents
Claude Code OpenAI Codex