Refactoring Orgs
Organizations accumulate debt the same way codebases do. Bad abstractions, unclear ownership, tangled dependencies. This is a blog about identifying that debt and fixing it.
I write about management, product, and the structural problems that slow down smart teams.
Writing
- Measuring Performance. Or How Much Performance Is Your Added Performance Costing You? They're probably 200% as fast as normal. And that costs us 50% of our performance on the backend. That number is not precise. It doesn't need to be. June 3, 2026
- "Every Token Is Profitable" Is the New "Every Seat Sold Is Active" Profitable tokens and valuable tokens are not the same set. The economics of production tell you nothing about the economics of value creation. June 2, 2026
- I Call Bullshit: The 20% Layoff at Cloudflare Is Not About AI Cutting 20% of headcount at a company posting record revenue is one of the most reliable ways to expand operating margins. The AI framing is just a cover story. May 21, 2026
- On KPIs In The Age Of AI Measuring AI adoption by usage is like measuring the health of a business by how many meetings its employees attend. The metric is real. The correlation is not. May 20, 2026
- Why Your AI Projects Are Failing (And It's Not the Technology) The bottleneck isn't code generation speed. It's decision quality upstream of code generation. Most teams haven't asked the right question yet. May 19, 2026
- The End of Onboarding As We Know It As teams generate more code than they read, nobody on the team knows why the codebase is the way it is. That breaks onboarding at its foundation. May 18, 2026
- Make Your Development Team Twice as Fast and You Ship the Wrong Thing Twice as Often Speed without alignment is a precision instrument for waste. AI removes the friction that used to catch wrong problems. That's not progress. May 17, 2026
- That AI Infrastructure You're Building Is A Red Flag For A Prioritization Problem The real problem is simpler: they have not decided which single workflow to fix first, and they haven't found out how to make money doing it. May 16, 2026
- Every AI Company Promising Democratization Is Building the Opposite Nobody knows what AI will cost in 2029. One thing is certain: "democratization" is not a safe planning assumption. Build your forecasts accordingly. May 15, 2026
- AI Access Is a Supply Chain Risk and Nobody Has Put It in the Risk Register Geopolitically sensitive resources get weaponized faster than organizations can react. Frontier AI access is in that category. The risk register doesn't know it yet. May 15, 2026
- You're Managing AI Like It's Software. It Isn't. The marginal cost of serving another user with a frontier model is not approaching zero. Your roadmap was built assuming it would. May 15, 2026
- The Org Chart Is Designed to Create the Misunderstandings You Then Hold Meetings to Fix Deep specialization produces better outcomes within a domain. It also makes people systematically unable to understand each other. The org chart made that a feature. April 20, 2026
- The Developers Getting the Most Out of AI Write More Explicit Code Than They Did Before The effective pattern is not: describe vaguely, accept whatever arrives. It is: specify precisely everything except the syntax. April 17, 2026
- A Faster Feature Factory Is Still a Factory That Builds the Wrong Things AI accelerates the build cycle on right and wrong things equally. An organization that systematically builds features nobody uses will do so faster with AI. The waste rate goes up proportionally. April 16, 2026
- A Large Codebase Is Not a Moat. It's a Mortgage. Accumulated complexity was supposed to be what made incumbents hard to dislodge. It is also maintenance burden, coordination overhead, and organizational inertia. April 13, 2026
- Stop calling trigger-happiness agile The engineer who knew it wouldn't work was not in the room when the contract was signed. That's not a communication failure. It's a structural one. April 8, 2026
- Why RTO is a red flag for Organizational AI-readiness Going back to the office wasn't a culture decision. It was a statement about where information lives - and that statement determines whether AI can do anything useful for your organization. April 7, 2026
- Your AI Did Not Ship That Bug. You Did. Quality is still a choice. The tooling changed. The accountability did not. April 6, 2026
- Most Companies Are Flying Blind and Calling It Strategy Every management layer is a translation step, and every translation loses something. For the first time, there's an alternative. April 5, 2026
- Your Middle Management might just be an Apology for Bad Leadership Middle management wasn't invented for efficiency. It was invented to compensate for incompetent generals. Most organizations still work exactly this way. April 1, 2026
- Agility Made the Same Mistake It Set Out to Fight The promise was small steps and fast feedback. What we got was sprint plannings that run three hours and backlog grooming as a full-time job. February 27, 2026
About Me
Hi, I'm Bastian Wilhelms. I am the COO of sipgate, one of Europe's most unconventional tech companies with a headcount of roughly 280 people - and co-responsible for building and running a genuinely hyperproductive organization. Not the kind that talks about productivity. The kind that has actually set out to make it work, structurally, at scale, over many years.
Views are my own. None of it represents my employer's official position on anything. I'm a heavy AI user, and I do use AI for research, writing and editing my posts (and translation into German, since I write in English).
I founded my first startup in 1997 and have been thinking about how organizations work - and why they so often don't - for longer than most frameworks for fixing them have existed.
I am based in Düsseldorf, Germany. You can reach me via Bluesky or encrypted mail.