Code got cheap. Your org didn't.
From individual coding agent usage to an engineering operating model. Hands-on support for engineering leaders.
I keep seeing the same pattern. CTOs roll out Cursor licenses, but productivity barely moves. Teams experiment a bit, fall back to old habits, and nothing changes.
It's not that people avoid AI. Full resistance is almost gone. But there's still a massive gap between using AI and actually adopting agentic engineering.
Knowing that coding agents exist and changing how 30, 100 or 500 engineers work are two completely different problems. That's the gap I work in. And the answer isn't a tool list. It's how the organization works.
Two ways I come in
Most teams I meet already have the tools, and many already had an enablement workshop. People know how to prompt. The organization around them hasn't caught up.
So I do two things. A hands-on workshop, so a team actually learns to engineer with agents. And a transformation engagement, for when the tools are in and the org is still chaotic.
AI Engineering Enablement
Your engineers already use coding agents. But are they engineering with them?
This is not two days of slides. We build.
- The techniques, on safe groundPrompting, spec-driven development, loop engineering, agent workflows, the agentic SDLC. Everyone builds working software in hours instead of weeks, so the techniques land as experience, not as slides.
- Your own backlogOn day two your team brings real items: bugs, refactorings, the thing nobody had time for. They attack them on your codebase, with me in the room when something goes sideways.
- Habits that survive the week afterHard to argue with software you built yourself, running on your own laptop. That's what turns the sceptics, and what makes the new way of working stick once I leave.
You leave knowing when to delegate, how to structure and review agent work, and how to build reliable loops instead of blindly prompting.
AI Engineering Transformation
Everyone has a coding agent. Individual engineers are faster. The organization is still chaotic.
I come onsite, talk to the people, and look at how the organization actually works. Then we decide what to do about it.
- ListenI talk to your people and look at the current process. How coding agents are really used across teams, where communication breaks, where knowledge doesn't move, and which problems people have stopped putting on the table.
- Problems and solutionsI come back with a list of what's broken and a set of potential solutions. Actionable enough that leadership can decide. Not a maturity model.
- RoadmapTogether with your leadership, we turn that into an execution plan: what to run first, who owns it, and what we check.
Then every two weeks I come back and we check the plan.
I've done this from the inside
30 two-day workshops this year, on real codebases, with real backlogs and real resistance. Teams at bonprix, IONOS, Thomann, CTS EVENTIM, Trusted Shops, Russmedia, Witt and Enpal.
I've hosted two CTO Bootcamps on the organizational side of this. A third one is coming in October in Hamburg. In November I'm speaking at Øredev.
I write about agentic engineering every week on LinkedIn. And I run a public comparison of coding agents and harnesses at agentic-engineers.dev. Everything I teach, I publish first.
Why an outsider helps
I'm Fabian Wesner. I've been all-in on coding agents for more than 1.5 years. Long enough to know which parts of the promise hold, and which parts quietly fall apart once more than ten people are involved.
A CTO gets pulled into the operational detail. I don't. And because I work across many engineering organizations at once, I see which patterns are genuinely emerging in a market that changes every few months, instead of the ones a vendor is selling this quarter.
Let's talk
30 minutes, no pitch. I'll tell you honestly whether I can help.