When an Engineer Asks an AI Who Has an Obsolete Part

An engineer at a medical device maker is three weeks from a production run. One part on the board, a 15-year-old serial EEPROM, is gone from every franchised distributor. She’s not a buyer, and she doesn’t know the independent market. So she opens an AI assistant and types: who sells obsolete memory parts, and which of them test what they sell?

The assistant writes back a short paragraph. It names a few companies, describes what each does, and mentions a certification or two. Maybe she calls one of them. Maybe she forwards the list to purchasing.

Your company has handled that exact part family for a decade. Whether your name made that paragraph depends on things you probably haven’t looked at in years: what your website says, whether your services are written down anywhere, and whether your company facts match across the places that list you.

I’ll be honest about where my knowledge stops. I’ve never sourced a component, and I won’t pretend to know your lines. I ran a digital agency, Infographic World, for almost 18 years, and now I use AI to fix real problems inside businesses. What I do understand well is how information about a company gets read, by people and now by software, and why a company that’s great at its work can still be hard to describe from the outside.

What an assistant has to work with

An AI assistant answering that question isn’t checking your inventory. It’s reading what it can find about companies that fit the question, then summarizing. That means the assistant can only say what’s been written down somewhere it can read.

For a lot of independent distributors, very little has been. The website has a homepage, a line card or a “request a quote” form, and an About page that says the company has been in business since a certain year. The real expertise (which families you know, how you test, which customers you serve) lives in the heads of the sales team.

So when an assistant goes looking, it finds a company name and a phone number. It can’t describe what it can’t read. A competitor with a clear page titled “Obsolete and end-of-life memory” is simply easier to describe.

Google has said publicly that there are no additional requirements to appear in AI Overviews or AI Mode, and that ordinary search best practices still apply, including keeping important content in text form. That’s useful to hear. It means the work here isn’t a trick. It’s the same clear writing that helps any buyer who lands on your site.

What makes a distributor easy to describe

I’d look at five things, in this order.

Clear pages for what you actually do

One page per service, written for the person who needs it. Obsolete and end-of-life parts. Excess inventory buying. Testing and inspection. BOM shortage support. Each page says who it’s for, what you handle, how the process works and what the customer gets back. Plain nouns, real part families, no slogans.

Product families named specifically

A line claiming every major brand tells an assistant nothing. A page that names the families you know well (memory, FPGAs, power management, military-grade parts) and says something real about each one, gives it something to match a question against.

Quality and certifications stated plainly

If you hold AS6081, ISO 9001 or AS9120, say so in a sentence, with the certifying body and where a customer can verify it. If your testing is done in-house or by outside labs, say which. Buyers worried about counterfeits ask exactly these questions, and an assistant can only repeat an answer you’ve published.

Consistent company facts everywhere

Same name, address, phone, founding year and service description on your site, your directory listings and your marketplace profiles. Inconsistent facts make a company harder to describe with confidence. I’ve covered this in more depth in one set of facts for local search and AI, and the general mechanics in how AI assistants choose local businesses. The same ideas apply to a distributor, even one that ships worldwide.

Real content that shows you know the trade

Articles that explain how you think about trace and testing show expertise in a way a line card can’t. If you want examples of the kind of thing that reads as real knowledge, look at trace grade as a price, what a 5962 number tells a buyer and excess and obsolete stock on the shelf. Written by someone at your company, in your own words, that kind of page does double duty: buyers read it, and so can assistants.

Where AI helps you do this

Most distributors don’t have a writer on staff, and that’s the real reason the site is thin. AI can close that gap without putting words in your mouth.

• It can interview your best salesperson by turning a recorded conversation into a draft page about the families they know.
• It can read your quote history and list the families and services you actually sell most, so the pages match your business rather than your aspirations.
• It can check your company facts across every listing you name and flag the mismatches.
• It can draft answers to the questions buyers ask your sales team every week, for your team to correct.

A person who knows the parts reviews every word before it goes up. A page that’s wrong about testing or certification is worse than no page.

Run the numbers

The formula in words: extra qualified inquiries per week, times the share you can quote, times your win rate on those quotes, times the average gross margin on a first order, times the weeks in a year. Then think about how often a new customer comes back.

Say, for the sake of the math, with numbers I made up:

• Clearer pages and consistent facts bring in one extra qualified direct inquiry a week, from any source, AI assistants included.
• You can quote 70 percent of them.
• You win 30 percent of what you quote.
• The average first order carries $1,200 in gross margin.

That’s 50 weeks × 1 × 0.7 × 0.3 × $1,200, or $12,600 a year from first orders alone. If half of those new customers order again a few times a year, the real number is a good deal higher.

To run it for real, pull three numbers from your own business:

1. Direct inquiries per week today. Count the ones that came to your site or email directly, not through the marketplaces.
2. Quote and win rate on direct inquiries. These often run higher than marketplace RFQs, because the buyer chose you.
3. Average gross margin on a first order, plus how many first-time buyers ordered again within a year.

What has to be in place first

• A website you can edit without waiting weeks on an outside developer.
• An honest list of what you do well. Families, services and certifications you can back up today.
• Certificates on hand with current dates, so every claim on the site can be checked.
• A list of every place your company is listed, with the facts each one shows.
• One person who owns the pages and reviews them twice a year.

Guardrails worth keeping

• No promises. No one can guarantee an assistant will name you, and your pages shouldn’t hint that anyone can.
• No claims you can’t prove. Especially on testing, trace and certifications.
• Don’t publish customer names without written permission.
• Don’t publish stock you don’t have. Pages describe what you do. Stock belongs in your real-time channels.

How you’ll know it’s working

• Direct inquiries per week, before and after the pages go up.
• Inquiries that mention a specific service page or family.
• New customers who found you through search or an AI assistant. Add the question to your intake form and ask.
• Whether a handful of buyer-style questions, asked of two or three assistants once a month, describe your company accurately. Answers vary, so watch the trend, not one result.

Where I’d start

Write one page. Pick the service or family you’re best at, sit with the person who knows it, and turn that conversation into a clear page with your certifications stated plainly. Then fix your company facts everywhere you’re listed. That’s a week of work, and it’s useful whether or not an assistant ever reads it.

If you’d like to see what your own request stream says about where to focus, start with the free week on the Hunt page.