No training wheels

A year of testing tools, reading the hype, and working out what actually helps.

I have not posted anything in a long time.

I spent the year reading instead. Trying tools, trying techniques, and reading a great deal of fluff about AI along the way. So I will say the thing most of those posts will not. We are not there yet. I am not sure we get there.

That is not a complaint. It is just the honest state of it. The demos that ship a finished product in an hour are not shipping products. They are shipping the appearance of one. What we are actually doing, all of us, right now, is quieter than that. We are building the processes. We are onboarding AI into the way we work. We are testing the limits and deciding, one task at a time, what genuinely helps and what only looks impressive.

There is no guide for this. Nobody has done it before us.

I am writing this for two people. The one who is afraid to use it, afraid to test things, afraid to get it wrong. And the one on the other side, overwhelmed by the hype, convinced the finished product is one prompt away. They are stuck in opposite directions, but they are both stuck.

Here is the thing I keep coming back to. This is not a bike.

You do not learn AI the way you learn to ride, with training wheels and a careful first lap around the driveway. Starting small and cautious teaches you almost nothing here. Start big instead. Throw your most enormous idea at it and watch what happens. Prototype the product you have been carrying in your head. Use it to see a flow you have only imagined, to understand a UI better, to feel out a UX before you commit to it. Push it until it breaks, because the place it breaks is usually the thing you actually needed to learn.

What you should not do is sit back and wait. Wait for it to design your interface. Wait for it to build your app out of nothing. That is the fluff talking, and it is the fastest road to disappointment.

Do the big, messy testing, and something quiet happens. You start to know it. You learn how it really works, what it is good for, what it is hopeless at, how to hand it the complicated tasks and keep the ones that are yours. You stop asking it to be magic and start using it as a tool.

For me, that is where the real gift showed up. It gives me back time. Time to think about the parts of a problem I never used to reach, the details I always ran out of hours for. Used that way, it becomes the assistant I always wanted. Used everywhere, on everything, without judgment, it becomes noise.

So that is where I have landed after a year. Use it wisely. Not everywhere.

If you are curious what a year of testing actually looked like, it was Cursor, Claude, Obsidian, Perplexity, Gemini, Flow, ChatGPT, Napkin AI, and Google Labs. Some stayed. Some did not. That part is the point. You only find out by trying.

If you have questions, drop me a message. Happy to help.