My wife noticed it before I did. The dryer — brand new, not even a year old — kept turning out laundry that was still damp. Not soaked. Just damp enough that you'd run it again and wonder if you were losing your mind.
Five years ago I would've Googled it, or pulled up a YouTube video and hoped somebody's dryer problem looked like mine. Instead I opened Claude and just described what was happening, the way I'd describe it to a repair guy on the phone.
First suggestion: disconnect the dryer hose, run a load, see what happens. I did. Bone dry. So the hose was the problem — not the drum, not the heating element, not some $200 circuit board.
Second question: okay, that was the culprit, now what? Check that the vent hose is connected to the outside port and still venting outside. I got in the crawl space. Sure enough, it had come loose. Taped it back up, started the dryer and venting outside. Problem solved.
Then I asked one more thing, mostly out of curiosity: what would a technician have charged for that same call in my area? Somewhere around $300 to $350 — and most of that gap was just me being willing to get under the house.
That's the whole story. No prompt engineering course. No paid plan. Just a real problem and a plain description of it.
Turns out I wasn't doing anything unusual. Harvard Business Review runs an annual study — thousands of real, logged AI use cases pulled from how people actually talk about using AI, not how companies say they're using it. Their 2026 numbers just came out: troubleshooting is the second most common use case for AI, period. Behind only therapy and emotional support.
Second place. Ahead of coding. Ahead of writing. Ahead of basically everything the headlines lead with.
I bring that number up because most of what gets written about AI adoption is a business story — top-down rollouts, big transformation budgets, executives arguing about ROI. And a lot of that activity, according to the same study, produces marginal results so far. Meanwhile the actual second-most-common use, sitting underneath all that noise, is a guy in a crawl space with a roll of vent tape. That gap between the AI headlines and how people really use it says more than either one on its own.
So if you haven't started using AI yet, or you tried it once and it felt like a bigger deal than it should've been — here's the question worth asking: why does starting feel so much harder than fixing a dryer vent?
"Getting started" isn't a skills problem. It's a permission problem.
I use AI for troubleshooting at work too. A lot of that is Alteryx — I'll get stuck on the syntax for a formula, something that should work and doesn't. I describe what I'm trying to do, paste in the formula I've been using, and give it the error it's throwing. Copilot walks through it with me until we land on what's actually wrong. Even a couple years ago, that same problem meant a support forum search or digging through a reference doc, hoping someone else had hit the exact same wall. Now it's a back-and-forth with something that actually knows what I'm trying to do, because I told it.
In the last issue I gave a personal life example. I asked for a batch of sheet-pan recipe ideas. I didn't just ask for recipes. I told it who they were for and what I was actually looking for — quick weeknight meals, minimal cleanup, food my family will actually eat. That's the entire difference between a generic answer and a useful one. Not a better prompt. Just more context, delivered the way you'd brief a new coworker instead of quizzing a search engine.
I put together a one-page cheat sheet on exactly this — how to brief AI like you'd brief a new coworker instead of quizzing a search engine. Hit reply and I'll send it over.
None of that required training. It required describing a real situation clearly — which, twenty years into a career, isn't a new skill you have to go learn. It's one you've been using the whole time, just aimed somewhere else.
Just Get Started
Here's what I'd tell you if you asked me where to begin:
Open a free account. ChatGPT or Claude. Don't read a guide first. Bring it whatever's actually on your mind this week — a problem, a question, something you're curious about, an opinion piece you want pushed back on. It doesn't have to be broken to be worth bringing in.
Three places to start, using what you already have:
Something broken or annoying around the house. Like the dryer. Describe the symptom, not your guess at the diagnosis. Let it ask the follow-up questions — that's the part that feels different from a search engine.
Something you're curious about. A current event you want a second opinion on, an industry trend you keep hearing about but don't fully understand, a topic you'd normally spend an hour researching. Ask it to walk you through it, then push back if something doesn't sit right.
Something small and personal. Meal planning, a trip, a gift list. Low stakes — a good place to practice giving context before you try it on something that actually matters.
You already have the hardest skill this requires: knowing what you actually want to understand. That's what AI needs from you. The rest is just typing it out.
Quick caveat, because I'd rather tell you straight than oversell it: this isn't a replacement for real expertise, and it doesn't turn every job into a DIY project. A dryer vent is a reasonable thing to tackle yourself. Your electrical panel or a real medical symptom isn't — AI can help you understand what's happening and ask a professional better questions, but it's not the professional. Know which kind of problem you're looking at before you decide how far to take it yourself.
One more thing, because I know the question's coming: when's it worth paying? Not because the free plan can't handle any of the above — it can. It's worth it once you want AI to remember context across conversations instead of starting over every time. I keep a dedicated Claude Cowork project on the Pro tier for exactly that. That's the point where a paid plan earns its cost. Until then, free isn't the training-wheels version. It's the actual on-ramp. It's what I used on the dryer.
One Thing to Try This Week
Pick one thing you're actually thinking about this week — doesn't have to be broken, doesn't have to be at home or at work. Bring it to AI the way you'd bring it up with a colleague, not the way you'd type it into Google. See what happens.
Hit reply and tell me what you tried. I read every response.
I know a chunk of you have been sitting this out, waiting to feel "ready." You already are. You just haven't opened it up yet.
See you next Monday.
— Eric