Launching a platform where verified experts publish content designed to be cited and named by AI systems.
🎉 Today I get to announce my latest product, Markwright.app. A verified AI citation engine for human expertise. After 3 months of building, it's alive! One statistic kicked this whole project off, 61.7% of AI citations right now are ghost citations. The link…
A single unapproved account can leak your full repository history including deleted secrets, and vendors can disable access remotely without your knowledge.
Someone on your IT team has a paid Grok account. You do not know about it. That is not a policy failure, it is just Tuesday. No attacker. No vulnerability. Nobody broke a rule. An engineer installed a CLI, pointed it at a repo, and the whole thing left the bu…
Building a system that improves itself means solving the human problem of delegation, not the technical one.
🔗 I had one AI hire another AI. Then I spent a week coaching the first one on how to be a better manager. The finale (for now) of my agent series is up. My system could run but could not improve itself. That made me the bottleneck, the exact single point of…
How shared memory architecture solved the coordination problem in multi-agent systems.
I love the British Library. Not just the building. The Librarians. So I built a Librarian agent who curates my full nine agent suite. A great librarian does something quietly astonishing. They hold the whole collection in their head, know what connects to wha…
One person, nine AI agents handling back office work, ten dollars a month in infrastructure.
🤖 I built a nine-agent company on a refurbished mini PC. It costs me about ten dollars a month to run. Nine AI agents handle the back office. They research, draft my content, watch my inbox, pitch me product ideas, and supervise each other while I sleep. All…
A video series on organizational alternatives for founders who don't fit conventional frameworks.
I spent years trying to fit a mold that was never built for someone who looked like me. This Pride, I stopped. I'm proud to share the first of three videos in a series I've been building. It's the work I most wanted to make this year, and I get to put it into…
A framework for splitting deterministic scoring from AI-powered analysis to control costs and risk.
When we built the MACH & AI Readiness Assessment at Fidget Labs, we made a deliberate choice. The scoring engine is 100% deterministic. Mathematically locked. No AI involved. No token cost. No hallucination risk. AI only shows up when it's time to generate yo…
Testing Gemini, ChatGPT, and Claude with the same question revealed different attitudes toward risk, reasoning, and user autonomy.
♨️ Last week I was setting up a portable AC in my attic room during a Dutch heatwave. The manual was useless. QR code went nowhere. Installation guide skipped half the steps. So I asked three different AI models the same question about connecting it to existi…
Bake data cleanup into your AI rollout instead of delaying launch.
🔥 Hot Take: Your data is a mess, do your AI project anyways!! Don't hold back your AI project because of your data. Bake the cleanup into your AI roll-out! We've all got inherited data we need to clean up. The thing is if you go down the traditional ways of…