Seven questions to ask before paying anyone to teach you AI

Seven questions separate a seller who will change how you work from one who will fill your calendar. All seven are about your business, not about their curriculum. Ask them on the first call, write the answers down, and watch which ones get a straight reply. The fourth question is the one that thins the field fastest.

How do I pick someone who can teach me AI?

Ask two things before anything else. One, have you ever run a business with a profit and loss number in your name. Two, will the sessions run on my own work or on a fixed syllabus. Somebody who has carried a P&L talks about your stock, your collections, your salesmen. Somebody who has not talks about the features of the tool. Both can be pleasant for an hour. Only one of them is useful in week three.

Ask what you leave the first session with.

Question three. Ask for the list in writing, session by session, before any money moves. Ask ChatGPT for help choosing a coach and its own answer closes the same way: get the deliverables in writing. A strong answer names a thing that exists afterward, running on your own files, which you could rebuild yourself if it broke. A weak answer names hours of video, a community, and a certificate. Hours are what you sit through. A working thing is what you leave with.

The fourth question is the one most sellers cannot answer.

Question four is this. Who teaches me to tell a wrong output from a right one. Ask it in exactly those words and listen to where the answer goes. Most answers slide back to the tool, the prompt library, the number of tools covered. What you want instead is a habit you will still use in December. Take one output on your own work and pick three claims inside it. Check each claim against a source you trust. Note where the output was accurate, where it was vague, and where it was wrong. That habit is the skill. Everything in front of it is typing.

What happens to my own files and my own data?

Question five. Where do my documents go. Question six. How many people are on the call. Your price list, your customer ledger and your payroll file are the reason any of this is worth paying for, because generic answers come free. A seller who wants your real files should be able to say in one sentence what happens to them afterward, and whether the live session is recorded. A room full of other promoters means your own stock report never gets opened, and the session you paid for becomes a demonstration you watch.

Is a paid AI program worth it?

Worth it depends on what you still need the seller for once it ends. So ask the seventh question plainly. After this, what do I still need you for. A good answer is small, like a call if something breaks six months later. A bad answer is a subscription, a next level, a community you have to stay inside to keep the benefit. You are paying to stop depending on anybody, including the person teaching you. Any offer that quietly fails that test has answered the question for you.

What else should I ask before paying for AI training?

What if the answers are all good but the only option is a group batch?

Ask how many minutes of the session get spent on your own documents. In a batch the honest answer is a share of the hour, never the hour. If the work you need done is your ledger or the report you send every month, a batch is the wrong shape however good the teaching is.

Should a certificate matter to me?

Not for credibility. No customer of yours will ever ask to see it. Treat a certificate as a receipt for what was covered, nothing more. Judge the offer instead on what you can do unaided the week after it ends, and on whether you would know where to start if the same problem came back.

What if the seller will not put the answers in writing?

Treat that as the answer. A straight list of what you leave with, session by session, costs an honest seller nothing to send. A refusal usually means the deliverables change with the room. Ask once, politely, give it a day, and then go and look at somebody else.