How a Non-Coder Builds an AI Employee (Free Account)

Issue 02 · The AI employee

Back in July I opened a store, and the first thing on the shelf was free: a short guide to building yourself an AI employee. The store closed in September 2026 and the guide came down with it. Everything that was in it is below, in full. It fixes the single biggest mistake I see people make with AI, and most of them don't know they're making it.

Why your AI keeps giving you mush

Here's how most people use AI. Open the chat, type a question, get an answer, close the tab. Ask it to write a post for your business and you get something that could be about anyone's business. Generic, polite, useless. So you decide AI is overhyped, and you go back to using it like a search engine with better manners.

The problem isn't the AI. The problem is that every one of those chats starts from zero. It doesn't know what you sell, what it costs, who buys it, or how you talk. It has never heard of you. You're asking a total stranger to write in your voice, every single time, and then being disappointed when it sounds like a stranger.

The AI you use every day is a stranger with amnesia. An employee is briefed once and never forgets.

What an AI employee actually is

Not a new app. Not a subscription. Not the chat window you already use, either. The big AI tools have a feature most people scroll right past: a workspace you set up once and keep. Claude calls them Projects. ChatGPT has its own version. Free accounts get it.

You load that workspace with a one-page brief about your business, once. From then on, every conversation you open inside it starts already knowing everything in the brief. Your prices. Your policies. Your voice. What you'd never say to a customer. You stop introducing yourself to a stranger every morning, and start handing work to something that was at the meeting.

What it does on day one

Real tasks, the kind that eat your evenings:

→	Customer emails. Paste the message in, get a reply drafted in your voice, with your actual prices and your actual policies. You read it, tweak a line, send.
→	The posts you never get to. "Three posts about the spring special" comes back about YOUR spring special, because it knows what's in it and what it costs.
→	The awkward ones. "Someone's asking for a refund on something we don't refund, how do I say that without losing them?" It knows your policy. It drafts the kind answer you'd have written on a better day.
→	Descriptions, quotes, FAQs, the website copy you've been putting off. All of it grounded in the brief instead of invented.

And the honest part, because someone should say it: it doesn't send anything, book anything, or talk to a customer on its own. Everything passes through your hands before it goes anywhere. That's not a limitation. For a one-person business, that's exactly the employee you want.

The moment it clicks

When I tested this, I built an employee for a made-up dog grooming shop, filled in the one-page brief, and opened a fresh chat. First question, cold: "What does a nail trim cost?" It answered with the exact price from the brief and added that walk-ins end at 4pm. Nobody had typed either of those things into that chat. It just knew, because it was hired, not searched. That's the whole thing, and with your real business it takes about thirty minutes.

How to do it tonight

No download, nothing to install, and a free account is enough. Open Claude or ChatGPT, make a Project, and put one page in it:

What the business is, in two sentences.
What you sell and what it costs.
Who buys it and what they usually ask.
Your policies, including the awkward ones.
How you talk, and what you would never say.

That page is the whole trick. Everything above comes out of it. If you know someone still treating AI like a search engine, this is the issue to send them.

Talk soon,
Tim

Some guy. Not an expert. Just building.

The AI you use every day is a stranger with amnesia

Keep Reading

Some Guy & AI