
My musings on anything from leadership to technology to entrepreneurship and back.
Two years ago, running an LLM on my own laptop felt like magic - but it was far too slow to be useful. Two years, a few chip generations, and a lot of model iterations later, the game has completely changed. With an M5 Mac and 64GB of RAM, I now run 27B and 35B-class models that are fast, capable, and good enough for real agentic work - including driving Claude Code with a local model. Here is how to get there.
There are two things that still give a modern engineer a real 'wow' moment. One is running an LLM locally. The other - the one this post is about - is writing your own agent: a small program that thinks, reasons, picks tools, and solves a real task on its own. I built one in Go, with zero dependencies, in a few hundred lines. Here is how it works.
AI transformation is not one project - it is three different games played at three different levels: the organization, the teams, and the product. Companies that treat it as a single initiative usually win at none of them. Here is how to think about each level, what works, and where the traps are.
When does hiring an Interim or Fractional CTO make sense? Seven real scenarios where temporary technology leadership delivers outsized value -- from leadership gaps to PE post-acquisition execution.
I had a silly idea: what if my homepage had a hidden arcade game where the actual text on the page becomes the enemies you shoot? I built it with Claude Code in about 30 minutes across three iterations. Here is how it went, what worked, and where I had to step in as the engineer.
Steve Jobs is the default template whenever leadership in tech comes up. But Bill Gates, Elon Musk, and Satya Nadella ran companies of comparable scale with very different styles. Looking at all four side by side is a much more useful lens than picking one as a hero - and the leader most managers should actually copy is probably not the one you'd expect.
The Stoics ran through plagues, civil wars, and political chaos while staying useful and sane. That is almost exactly the job description of an interim CTO. Here are seven lessons I keep coming back to.
Practical lessons from introducing AI-accelerated engineering in real companies. How agentic coding, automated testing, and AI-driven migration are changing how software gets built -- and what leadership needs to do differently.
When your top customers dictate your roadmap, your product becomes five bespoke solutions wearing a trenchcoat. Let's discuss a practical framework for shifting from customer-dictated to product-driven without losing revenue.
The Performance Improvement Plan is one of the most important tools for managers. Done right, it helps the person, the team, and the organization.