AI
The Pipeline Didn't Catch a Bug — It Invented Twelve
I moved a working local build into a CI/CD pipeline to test my own thesis. Every failure was an artifact of the pipeline, not the code.
Building a Self-Hosted Blog-to-Audio Pipeline on a $150 GPU
No cloud APIs, no monthly fees. A fully automated pipeline from Ghost post to voice-cloned audio narration, running on a used RX 580 in an LXC container.
Shifting Constraints
Constraints never disappear, they just shift where they show up. AI has torn down where we used to expect them, and where they're showing up may be the difference between teams succeeding or merely keeping up.
The Foundation Still Matters
AI tools amplify whatever you bring to the table. The eight thousand startups rebuilding their apps right now brought the wrong thing.
The Bus is Leaving...
Government agencies are spending years deploying yesterday's AI tools while agentic development compresses delivery from months to days. The gap isn't linear anymore. It's exponential.
Open Source Is Struggling and Open Source Might Be the Answer
The social construct is fraying. The principle is thriving. The distinction between the two is the only thing that matters.
WASM Isn't Always Faster
WebAssembly delivers on its promise, but the promise is more specific than the marketing suggests. The advantage isn't raw speed — it's predictable performance in the face of computational unpredictability.
Deploy From Here
The centralized CI/CD pipeline was the right bridge for its era. Containers, powerful workstations, and agentic tooling put us on the other side.
Token Spend is a Vanity Metric
The companies pulling back aren't proving AI doesn't work. They're proving that vanity metrics and unlimited budgets are a bad combination regardless of the technology.
Mythos vs. the Monolith
Anthropic announced recently that their new Mythos model is too dangerous to release to the public. It can find vulnerabilities and exploit them at a rate never before seen. And while many of us are taking this news with a grain of salt since all the AI companies are in a hype race as much as they’r
Your AI strategy is probably too complicated
AI continues to consume all the oxygen in the room. We’ve gone from surprise to memes to denial with bouts of anger and sadness. And maybe just a touch of nihilism as we watch the multi-trillion dollar companies in this space maneuver, collapse entire consumer ecosystems, and appear to be rushing fa
The Claude Code Leak: A Masterclass in the "Iceberg" of Engineering
Claude Code leaked recently due to a misconfigured software pipeline—the digital equivalent of leaving the back door to the vault wide open. The internet reaction was swift; while it wasn't quite a "Gangnam Style" server-breaking event, people rushed to grab the source code before the gap was plugge
The AI x Workflow Conundrum
We love talking about AI in extremes; we focus on total disruption, mass unemployment, or the rise of robot overlords. We spend endless cycles on financing and existential risks, often overlooking the practical reality of how this technology actually shows up at work today.
Peak Software?
Back in the 1950s, a paper was published suggesting that there was an upper limit to the amount of oil that could be produced globally, which was expected to be reached in the early 2000s. After that point, it would all be used up, and oil as a resource would slowly and surely be exhausted, and with
Markdown is the new way to use text, change my mind.
If you don't know what Markdown is, you might be in the same position as someone 40 years ago who was still struggling to figure out what this WordPerfect thing was and how it was better than the IBM Selectric.
Key Principles for Developing with AI
The article was originally posted on LinkedIn. During earlier experiments, I discovered key guidance for framing prompts and conversations when using AI as a "pair programmer" or "tasked agent." Extensive research exists on how to optimize model output, with various approaches like RISEN offering di