Blog
Practical lessons from running AI in a real business: agents, governance, engineering, and working with ADHD.
102 published posts
Browse topics and series
No posts match these filters. Try another word or choose another topic.
Latest posts
What actually changes when I resend a prompt that just failed?
Less than you’d think. A retry is a brand-new prompt, not a second attempt at the same one — the model has no built-in memory that "this exact request just failed," so unless you or the surrounding system change...
What does my prompt inherit from a long-running conversation?
Everything that came before me, whether it’s still relevant or not. By the time I’m the fortieth prompt in a long thread, I don’t arrive at the model alone — I arrive dragging the full history of everything asked and...
What happens to my prompt before the model actually sees it?
Your typed words are the smallest part of me by the time I reach the model. Before I’m ever processed, a system assembles me: it stitches your message onto standing instructions, relevant files, tool definitions, and...
What Silently Resets an AI Context Cache? The Exact Rules of Cache Hits and Misses
An AI context cache resets whenever anything earlier in the prompt changes — and "anything" includes settings you might not think of as part of the prompt: the reasoning-effort level, the tool list, the system...
Why Can’t My AI Agent Reach Its Tools? (The Same Fix I’ve Used for 25 Years)
Short answer: it’s almost certainly DNS, or the modern equivalent of it — a name, endpoint, or hostname your agent can’t resolve or reach. I’ve been in this industry since 1988, and in the last twenty-five years or so,...
Why did my prompt take longer to answer when the model used a tool?
Because answering me wasn’t one step, it was several. When a model can’t answer from what it already knows, it pauses mid-turn, calls a tool — a search, a file read, a command — reads the result, and then folds that...
Why did my repeated prompt come back faster and cheaper the second time?
Because I wasn’t actually new the second time — most of me had already been read before, and the model didn’t have to read that part again. When a large share of what I carry is identical to a prompt sent moments...
Why does a vague prompt cost more than a specific one?
Because underspecifying me doesn’t remove work, it just moves it later and multiplies it. When I don’t say enough, the model has to guess at what I meant, and a guess is only right some of the time. Guess wrong and you...
Why does the model already know things I never told it in my prompt?
Because you’re not the only one talking to the model. Before your message ever arrives, a separate set of standing instructions — the system prompt — has already told the model who it is, what it’s allowed to do, and...
Why I Still Build My Own Lab in the AI Era (And Why You Should Too)
I started in IT in 1988. Somewhere around 2011, I spent a year on the road running all-day TechNet events teaching IT pros how to build their own private cloud — Hyper-V, System Center Virtual Machine Manager, System...
Why should I send all my questions in one prompt instead of one at a time?
Because the master orchestrator—your main session chat—can see the whole working context in one call in the current live conversation. Put new questions, your responses to earlier answers, and the next work in one...
Why your AI can’t be down when the engine is open
A technician has an engine open, hands full, and a question that needs an answer right now: is this reading normal, or is it the reason the equipment is about to fail. They ask out loud, because their hands are not free...
Why Can a Second AI Opinion Still Leave You With the Wrong Decision?
Ask a second AI for a check on a plan, get an answer that agrees with the first one, and you still have no idea whether the work is safe to fund. In our own platform planning, the working plan said 8 features and 24...
Why Isn’t a Merged Pull Request the Same as an Approval?
Pick one of the last five consequential changes we merged. If we cannot show who authorized the next action for those exact bytes, under which current policy, and with what current evidence, we are one deployment away...
Is the Rule You Rely On Actually a Control, or Just a Wish?
Pick one consequential rule in an AI-assisted workflow. List every route that can perform the action and the current denial evidence for each route.
Why Did My AI Forget What I Taught It Yesterday?
Your AI appeared to forget what you taught it yesterday because the later session did not retrieve the decision that should govern its work. That can happen even when a product saves conversation history or maintains...
Should an AI Verifier Agent Check a Bug Diagnosis Before the Fix Author Starts?
Of seven diagnoses in one documented pass, independent verification overturned two and deepened one. The same pass also caught a non-persistent proposed repair before merge. Those outcomes put fix-author time, AI spend,...
Why Does One Bug Filed as Two Tickets Get Fixed Twice?
If one bug appears in two tickets, can we prove before work starts that our people and AI agents will not repair the same outcome twice? Without that proof, the second authorization can repeat staff hours, AI calls,...
