What to bring to a first conversation about AI

Operations managers often put off a first conversation about AI because they feel they ought to arrive prepared. Prepared, in their minds, means a list of what they want the AI to do, a view on which model to use and someone technical in the room. None of that is needed, and waiting for it is the main reason the conversation never happens.

What a first conversation needs is something you already have in full: a clear account of how the work is done today. This article is about what that account looks like, what a good first hour feels like, and what you should expect to get back.

You do not need a wish list

A wish list written before anyone has looked at the floor is a guess. It tends to be shaped by what other companies say they are doing, or by a demo someone saw at a trade show, rather than by where your own people lose time. If you arrive with one, we will put it to one side politely and come back to it at the end.

You also do not need technical knowledge. Nobody on your side has to know what an API is (the doorway other software uses to talk to a system), which AI provider is better, or whether your database is suitable. Those are our questions to answer, and we answer them by looking, not by asking you.

Bring how things work today

The useful material is an honest description of a normal shift, in the words of the people who live it. Four things make that description complete.

• Who does what. The receiving clerk, the dock coordinator, the supervisor on the afternoon shift, the person who answers client emails. Names are not needed; roles and handovers are.
• Which screens and paperwork. The warehouse system, the spreadsheet that acts as a dock calendar, the shared mailbox, the printed pick lists, the scale tickets, the delivery notes that get photographed. Screenshots and a few blank forms are worth more than any diagram.
• Where people wait. The moment a coordinator cannot plan the afternoon until a carrier confirms. The supervisor who waits for a stock count before releasing a wave. The clerk who cannot enter an order until someone finds the missing weight.
• Where people re-key or re-check. Information that already exists somewhere and gets typed again into another system, and the places where a person compares two sources by hand because one of them is sometimes wrong.

If a few people on your side each write half a page on their own part of the shift, you have more than enough. It does not need to be tidy. The messy version is more useful, because the detours are where AI fits.

What a good first hour looks like

A good first hour is mostly you talking. We ask what happens from the moment a truck is booked to the moment the last pallet is put away, and we keep asking "and then what?" until we reach the end of the shift. We write down every point where a person waits, re-keys or re-checks, and we ask how often it happens and what goes wrong when it is done late.

We do not sell during that hour, and we do not propose an integration before we have heard the whole shift. If a candidate is obvious by the end, we say so and explain why it stands out. If nothing stands out yet, we say that too, and we ask to walk the floor during a normal shift before anyone decides anything.

One contact on your side is enough to start. A supervisor or an operations manager who knows the floor is ideal; an IT person is welcome but not required for the first conversation. We will need IT later, when we look at how the integration will reach your systems.

What you get back

After the first conversation, we come back with a short list of candidates rather than a proposal. Each one names the task, who does it today, the data it would need and where that data lives, and whether it would run in the cloud or at the site. We also say which one we would start with and why, usually the one where the re-keying or waiting is most frequent and the data is already reachable.

From there the steps are the ones we always follow: we agree the first integration together, redfly designs it, we review the design with your supervisors, redfly builds it, a small group pilots it on real work, it is tested against a set of checks, and it is rolled out with our team watching.

Bring the shift as it really runs; the list of what AI should do comes out of that, not the other way round.

This is how redfly begins with every warehouse and distribution company: you bring how the operation runs, we work out where AI fits, and we build that first integration into the systems you already run, with the results landing in the redfly Portal for your people to act on.