UNFAIR
DISTRIBUTION

Help while the purchase is taking shape

AI assistants can bring suppliers into a purchase before the design is settled. Helping then may earn business, but the cost includes requests that never become orders.

Two small figures discuss a red part design at a counter while a third waits with another drawing; a folded part stands nearby.

I wanted to put an ultra-short-throw projector in my media room. The console wasn't deep enough, and the room had other constraints to work around. Astra, the AI I was working with, helped me work through the measurements. We came up with an aluminum cantilever shelf that would wrap around the console, and Astra operated Autodesk Fusion on my Mac to create the model and files for fabrication quotes.

Those files went by email to three metal shops. When I posted about it, one had replied and Astra was following up on the finish.1 I had a proposed shelf and a quotation conversation.

A flange that's too short is one problem a shop might find in a design like this. That's the straight material between a bend and the sheet's edge. In a separate example, the manufacturer Fabworks documents an agent quoting service that can flag this problem and return dimensions and limits for the selected material.23 An assistant can use that information to identify what needs changing, attempt a revision, and ask for another check.

A design that fits the room may need changes before a shop can make it, and those changes may affect whether it still fits the room. The supplier's production knowledge can help determine what the customer ends up asking to buy. For a small business, I think there's a distribution opportunity in making that knowledge available while the request is still taking shape. You get a reason to enter the conversation before the customer has reduced the purchase to a specification and a price comparison.

Being useful early still leaves the customer free to buy somewhere else. A revised design can go to another shop, just as Astra sent the files for our project to three shops. The same assistance that gets a customer closer to buying can leave them better equipped to shop around. That makes this an uncomfortable place to offer unlimited help: your team can spend time on a request even when the customer leaves, or never buys anything at all.

If I were deciding what to build for a small business, I'd start with the questions that establish whether we can fulfill a request. A production limit belongs there because it tells the customer something about what we can deliver. Broader design work can keep expanding after that question is answered. The distinction gives a team a way to set the scope of its assistance, and to decide where a person needs to take over. A manufacturing check, for example, doesn't establish that a shelf is safe for its intended use. Offering the check shouldn't imply that someone has taken responsibility for that separate judgment.

The scope matters commercially even if the assistant does most of the work. Fabworks lets an agent prepare and revise a quote, but the customer reviews it and completes checkout and payment in the dashboard.3 A request that makes it through preparation can still stop there. I'd judge an investment in this kind of help by the paid work it brings in relative to the cost of supporting all the requests, including the ones that go nowhere. Charging for preparation is another possible choice, though it changes the offer and may change how many customers want the help. A small team needs room to discover that before it promises to work through every unfinished idea for free.

BUILD A WAY TO
YOUR NEXT CUSTOMER.

Unfair Distribution is a practical book for founders who have to make the next sale count.

Get the book Read an excerpt