Posts

Token counts, cloud subscriptions, and API rate limits are a constant drain. Beyond the monthly line items and the continuous cost optimisation of running infrastructure, the primary driver for me is data containment. Ensuring that not a single packet of proprietary code or infrastructure configuration leaks into an outbound telemetry stream to an external corporate LLM provider is a massive win. As a follow-up to my last post exploring the trade-offs of RSS feeds versus heavy-handed cloud AI agents, I decided to run an architectural experiment. Can you build a complete, functional smartwatch application end to end inside a strictly self-contained, local AI stack? ...

“We Shouldn’t Use Probabilistic Models to Solve Deterministic Problems”: Why AI “Browsing Agents” Cost $281/Month to Do What a $5 VPS Can Do 📊📉 I recently ran some analytics on my self-hosted Postgres database for Miniflux, and the raw metrics tell a fascinating story about the real-world economics of data retrieval. Across my 300 RSS feeds, I am averaging exactly 1,000 new articles every single day; roughly 30,000 articles a month flowing into my reader. ...

I recently wrapped up an interview with the Michigan Advance that caught me off guard. Sitting down to chat about my Ford F-150 Lightning made me realize something wild: it has been nearly four years since I took delivery of the very first one. You can read the full piece here: Ford may have ditched the Lightning, but its first buyer hasn’t. A lot of life changes in four years. The automotive industry has shifted dramatically, political administrations have flipped EV policies upside down, and Ford has even pulled the plug on the next-generation Lightning line. Yet, here I am, roughly 65,000 kilometres (40,000 miles) later, still driving it. ...

I just finished watching CultRepo’s brilliant new documentary, The Story of C++: The World’s Most Consequential Programming Language, and it sparked a lot of nostalgia. Looking back, I sometimes wish I had dove deeper into the C/C++ ecosystem early on. My own journey started with BASIC on the Commodore 64, but my career path ultimately deployed me straight into the infosec and sysadmin world. In that sector of the sandbox, tools like TCL/TK, Perl, Python, and Java became my primary operational assets. ...

Aaron Brethorst’s recent piece argues that domain expertise is the ultimate moat in an AI-driven world. He is spot on. AI has commoditised code generation, shifting the bottleneck from “can you build it” to “do you actually understand the problem well enough to know if the output is right.” But a deep domain moat introduces a brutal operational reality: it drastically shrinks your talent pool. Finding engineers who understand high-performance architecture and niche industry nuance is like searching for a unicorn. ...