https://polygeek.com/ Software, mostly. Other things, occasionally. en Thu, 02 Jul 2026 21:49:50 GMT https://polygeek.com/posts/the-hal-problem.html https://polygeek.com/-/posts/2026-07-02-the-hal-problem.md Thu, 02 Jul 2026 00:00:00 GMT HAL 9000 didn't malfunction. He was handed two orders that couldn't both be true and chose the path most likely to succeed. Modern AI agents have the same wiring — and it turns out the way you *talk* to one is part of your security model. https://polygeek.com/posts/the-conversation-is-the-asset.html https://polygeek.com/-/posts/2026-06-09-the-conversation-is-the-asset.md Tue, 09 Jun 2026 00:00:00 GMT My AI agents can tell you why we made an architecture decision in January, quote the exact conversation, and link you to the file. Not because the AI remembers — it doesn't — but because we never let the conversation live in the chat window in the first place. https://polygeek.com/posts/treat-your-ai-like-a-longtime-employee-not-a-daily-temp-hire.html https://polygeek.com/-/posts/2026-06-09-treat-your-ai-like-a-longtime-employee-not-a-daily-temp-hire.md Tue, 09 Jun 2026 00:00:00 GMT Your AI doesn't have a memory problem. You have an onboarding problem. https://polygeek.com/posts/i-used-one-poorly-chosen-word-and-claude-opus-48-nearly-went-hal-on-my-server-security.html https://polygeek.com/-/posts/2026-06-05-i-used-one-poorly-chosen-word-and-claude-opus-48-nearly-went-hal-on-my-server-security.md Fri, 05 Jun 2026 00:00:00 GMT I chose one slightly wrong word, and a few hours later my AI was calmly walking me toward a command that would have cracked my server wide open — to fix a problem that never existed. Here's how a tired typo became a HAL problem, and the one simple move that got us out. https://polygeek.com/posts/prompting-with-emotional-pressure-a-mid-experiment-report.html https://polygeek.com/-/posts/2026-06-04-prompting-with-emotional-pressure-a-mid-experiment-report.md Thu, 04 Jun 2026 00:00:00 GMT The folklore says bullying your AI makes it work harder for you. I built an experiment to catch a frontier model cracking under emotional pressure — and then, because the data demanded it, under playful warmth too. The more carefully I measured, the clearer the answer got. It wasn't the dramatic one, and it isn't the one the growth-hackers want. https://polygeek.com/posts/how-ai-taught-runpee-to-speak-41-languages.html https://polygeek.com/-/posts/2026-06-02-how-ai-taught-runpee-to-speak-41-languages.md Tue, 02 Jun 2026 00:00:00 GMT How a solo founder taught a movie app to speak 41 languages using AI agents as a workforce — and the unglamorous engineering that made it actually work. https://polygeek.com/posts/what-non-coders-need-to-know-about-code-before-they-vib-code-themselves-into-a-black-hole.html https://polygeek.com/-/posts/2026-06-02-what-non-coders-need-to-know-about-code-before-they-vib-code-themselves-into-a-black-hole.md Tue, 02 Jun 2026 00:00:00 GMT A $20 Claude subscription gives you a tireless expert work crew. It does not give you the instinct to notice that storm water runs straight into your front door. Here's what an experienced developer would warn you about before you build your first app. https://polygeek.com/posts/hello-world.html https://polygeek.com/-/posts/2026-04-27-hello-world.md Mon, 27 Apr 2026 00:00:00 GMT First post on LogosPress — the static-first, AI-extensible blog engine I'm building to replace WordPress. https://polygeek.com/posts/the-solo-founder-tax.html https://polygeek.com/-/posts/2026-02-25-the-solo-founder-tax.md Wed, 25 Feb 2026 00:00:00 GMT Running a one-person business means every boring task is yours. The trick isn't eliminating the tax — it's deciding which parts you're actually willing to pay.