A standing hour, every week.

For manufacturers and distributors who know AI matters and want to find out what it can actually do here, in this building, on these jobs, before anyone builds anything.

What happens in the hour

One hour, same time every week, on video with screens shared, on whatever you're working on that week, whether that's a quote you're putting together, a report nobody can get out of the system, or a procedure that only lives in one person's head. You bring the work. There's no curriculum and there are no slides, and nothing about it is a class.

We open the real thing, not a sample, and we work on it together, and the skill that comes out of it is yours to use from then on.

An hour a week is where most of this starts. If you want more than that, or more of your people in it, it gets sized to what you're actually trying to get done. It isn't a package with a number on it.

Between sessions I keep going on whatever came out of it, trying an approach to see whether it holds up, or writing a skill for it, and you get it back at the next session. A skill is your way of doing a job, written down so the AI follows it instead of guessing at it.

What you're left with at the end of a month

Not a plan. Three things you can point at:

• The skills we build stay in your workspace and keep working whether I'm in the room or not. They're yours, and they don't stop when we stop.
• A short written summary each month of what we tried, what worked, and what didn't. The ones that didn't are worth as much as the ones that did, because that's the money you don't spend.
• Your own work, sorted three ways, and added to as we go: where a chatbot is enough on its own, where it needs your procedure written into it first, and the few things that need something built, because the answer has to be right every time.

Why do this with me

I've spent nearly 25 years in manufacturing, including a decade running a machine shop. And I build and run real software for manufacturers: a monitor that reads vendor acknowledgments and flags the POs that came back wrong, Customer PO intake that reads customer POs instead of somebody re-typing them into the system, operating reports that land in your inbox instead of waiting for somebody to pull them. So on the day something does turn out to need building, you're already talking to the person who would build it, and I'll tell you honestly when it isn't worth building at all.

That last part matters more than it sounds. AI reads, drafts and remembers what your people already know. It isn't exact and it doesn't run your systems. Knowing where that line sits, for your work specifically, is most of what the hour is for.

Where it fits

You can start here, and plenty of operations should. If you'd rather start with everything set up in one go, that's the paid half-day, $2,500, where your team's AI gets set up and your people learn to write skills. Either one can lead to the other, and neither one requires the other first.

Month to month. Stay as long as it's earning its keep, and step away the day it isn't.

Common questions

What does it cost?

It depends on the level you want, so I won't throw a number at you before I know what that is. We talk through what you're trying to get done and who needs to be in it, and I send you a price after that. From there it's a flat monthly, billed month to month, with no hourly meter and nothing to sign for a year.

How is this different from an AI course?

A course teaches you AI. This works on your jobs. There are good free AI courses funded nationally right now and I'd point you at them before I'd charge you for the same thing. What none of them do is open your quote, your report or your procedure and work on it with you.

What if it turns out we need something built?

Then you'll know that with a reason behind it instead of a hunch, and I build it. That's quoted on its own, up front, separate from the monthly. The monthly never turns into build hours, because that's how both things end up done badly.

Is our data safe in this?

It's set up inside the environment you already run, under your own accounts, and none of it is used to train the underlying AI models. The longer answer, including prints under NDA and export-controlled work, is here: is ChatGPT safe for company data.

We tried AI once and nobody used it.

Usually the reason has nothing to do with the people: why most AI projects fail in small manufacturing. A standing hour is what most of those rollouts never had.

Related reading: AI is a good reader, it's a bad accountant, on where AI stops and regular software starts. And capturing the know-how that walks out the door, which is the job most of these hours end up circling back to.