What to ask an AI vendor before sharing job data

A software salesperson walks the shop floor, or calls a trades business owner between jobs, with an artificial intelligence (AI) add-on that will read jobs, quotes and customer files. The demo may sound useful, but connecting it would give another company access to information the business has spent years collecting. Before that vendor touches any job data, the owner needs a short list of questions and clear answers in writing.

The goal is a record the owner can read, compare and keep. A useful answer names the places information goes, the people who can open it and the choices available at cancellation. A sales presentation cannot replace those answers.

What the checklist covers

The checklist follows one set of information from the shop to the vendor and back again. Start with a real example, such as one quote containing a customer name, address, job details and price. Ask the vendor to explain what happens to that exact record.

• Where does our data go? Ask which systems receive it, where copies are stored and whether any other company handles it. Request the same explanation for information uploaded during setup and information collected while the tool runs.
• Who can see it? Ask whether vendor employees, contractors or other companies can open job and customer records. Also ask what access the owner can restrict and whether the vendor will describe those limits in writing.
• Will our information be used for training? Training means using business information to improve an AI system. Ask whether jobs, quotes, customer details or employee entries can be used that way, and ask how the business can decline that use.
• Can we take our information with us? Ask how the owner can get a usable copy of the business's information before cancelling. The answer should explain what the owner receives and how to request it.
• What happens after cancellation? Ask when access ends, what happens to stored records and how the owner can request deletion. Get the cancellation steps and any conditions in the written agreement before signing.

Send these questions by email instead of relying on a call. Written answers make omissions easier to spot and give the owner something to compare with the contract. If a response introduces another unclear term, ask for a plain-English explanation using the same sample quote.

Who should use it

This checklist is for Capital Region machine shops and trades businesses considering any AI tool that could touch job data or customer information. It also helps office managers and other employees who receive the sales pitch but need owner approval before connecting company records. One person should collect the answers and bring unanswered questions back to the owner.

A business trying AI only with invented examples or public information has less job data at stake. The same questions become important as soon as real quotes, customer records or job files enter the picture. The owner can keep early experiments separate from live business information until the answers are satisfactory.

How to use the checklist before signing

The owner can repeat the same process with every vendor. That makes the answers easier to compare without requiring technical knowledge. It also gives the business a consistent point at which to pause a purchase.

Red flags that mean walk away

• The vendor will not explain where job and customer information goes.
• The vendor avoids naming which employees, contractors or other companies can see it.
• The vendor insists on using the information for training after the owner declines.
• The vendor cannot explain how the business gets a usable copy of its information.
• The cancellation or deletion answer changes depending on who is speaking.
• The vendor pressures the business to connect live data before providing written answers.

One unanswered question is a reason to pause. Repeated avoidance, a refusal to put answers in writing or pressure to skip the review are reasons to walk away. A shop does not need to accept vague answers just because the demonstration looked useful.