The ten-minute crossing

The gap between rung 0 and rung 1 is the largest one on this ladder, and it is almost entirely psychological. Nothing about it is difficult. It is presented as difficult, by people who benefit from it being presented that way.

Here is the whole crossing.

1. Download the desktop version of the AI assistant you already pay for.
2. Point it at one folder. Not your whole machine. One folder, with about twenty files in it.
3. Ask it a question about the contents that you already know the answer to.

That is it. Ten minutes, and you will have learned the only thing rung 1 teaches: the difference between an assistant you brief and an assistant that already read the file.

Why step three matters

Ask a question you already know the answer to. This is the part people skip, and it is the part that protects you.

The failure mode of AI at work is not that it refuses. It is that it answers confidently and wrongly, and you cannot tell, because if you could tell you would not have needed to ask. Asking something you can check turns the first ten minutes into a calibration exercise instead of an act of faith.

You are measuring two things. Did it actually read the files, or is it guessing from the filenames? And when it was wrong, was it wrong in a way you would have caught?

The objection, stated fairly

The most common reason people do not do this is not cost and not scepticism about AI. It is this, in their own words:

“I’m not technical.”

Underneath that is usually a specific memory rather than a general belief: an evening lost to an integration that never quite worked, or the fear of looking foolish in front of someone who does this for a living.

Which is why the version above has a hard stop built into it. One folder. One question. Ten minutes. If it has not worked in ten minutes, close it — the failure is the product’s, not yours, and the answer is a different tool rather than a longer evening.

What not to do next

Do not immediately connect everything. The reason to go one rung at a time is that each rung teaches you which work is worth handing over, and handing over the wrong work is how automation projects die.

Spend a fortnight at rung 1. Notice which questions you keep asking it. Those are the jobs worth automating later. Everything else is enthusiasm.