Most owners meet AI the same way. You hear it's going to change everything, you feel a little behind, so you open a chatbot and ask some version of "What can AI do for me?" You get a tidy, confident answer. It's also generic, because you handed it nothing to work with. So you close the tab and conclude the hype was overblown.
I did exactly this. The answer isn't a better chatbot or a cleverer question. It's that you asked too early.
Here's the reframe that changed how I run my company: AI is the first coworker that's only as good as what you've written down. A new hire who knows your business cold can act on a one-line request. A new hire who knows nothing needs everything spelled out — and even then guesses. AI is the second one, every single morning, unless you give it a memory. The work isn't prompting. The work is context.
The thing nobody wants to hear
The reason AI feels underwhelming in a hands-on business usually isn't the model. It's that the knowledge it would need lives in three places a machine can't reach: scattered files nobody can find, tribal knowledge in your best people's heads, and your own judgment that you've never actually written down.
That's not an AI problem. It's a documentation problem. AI just makes it impossible to keep ignoring, because the moment you try to delegate something real, you run straight into how little of your business is written down in a form anyone — human or machine — can use.
The good news: fixing that pays off whether or not you ever touch AI again. Documentation makes onboarding faster, makes you less of a single point of failure, and makes the business worth more if you ever sell it. AI just turns it from a chore you postpone into the thing that unlocks everything else.
What "context" actually means
Context isn't a 90-page manual. It's the handful of things you'd tell a sharp new hire on day one so they stop guessing:
What the business does, who the customers are, and how you actually make money.
How you talk — to customers, in bids, in the field. The words you use and the ones you'd never use.
Your standards and your hard-won rules. The mistakes you've learned not to repeat.
Where the real numbers and documents live.
Write that down once, keep it somewhere the AI can read, and every task starts from your business instead of from a blank, average internet.
How to start this week
You don't need software for this. You need an hour and a willingness to think out loud.
Brain-dump one process. Pick one thing you do over and over — building a bid, onboarding a tech, closing out a job. Talk or type everything you know about how it really gets done, mess and all. Don't organize it. Just get it out.
Make the AI interview you. Paste the dump into a chatbot and say: "Ask me questions until you could write a clean SOP for this. One question at a time." This is the trick most people miss — you don't have to be organized up front. Let it pull the knowledge out of your head.
Save the result as your first context file. A plain document. Your business, in your words. That file is now something you hand the AI before any related task — and something a new hire can read on day one.
Use it on a real task. Next bid, next policy, next training doc — give it the file first, then the request. Watch the output stop sounding like a stranger and start sounding like your company.
Do that three times and you'll have the start of a real operating memory. Do it for a year and you'll have what took me a year of doing it wrong to build.
The one line to carry out
Stop shopping for the perfect tool. The tool matters less than the vendors selling it want you to believe — and far less than the context you feed it. Write the business down first. Then pick your tools. Then iterate inside your own company until the thing holds up on a bad week.
You don't have an AI problem. You have a writing-things-down problem. The fix starts with one process and one hour.
Next week: the one-hour way to start when you feel hopelessly behind. If something here was useful, hit reply — I read everything.