Labs
- We tried to claim our business in ChatGPT on day one. The trades can't get in yet.
OpenAI opened business profiles in ChatGPT today. Storefronts and online stores can claim one. The roofer, the plumber and the window installer are marked coming soon.
Field notes
- An automated caller asked a building-products supplier for a window price
One reported call. The questions that followed were not about the technology. They were about who the caller worked for and whether anything it asked for should ever turn into an order.
Field notes
- Voice-powered content management
An owner says what changed. The website is updated, checked on the live page, and the owner is told when it is done.
Systems
- Your company's most valuable knowledge may not be in its software
Most companies can list their software. Few have written down how the work actually gets done. AI is about to make that gap expensive.
Research
- Audit Lab
Audit Lab is a website check that ties every finding to something it measured, and says could not check when it could not. In October 2026 one report broke that rule.
Experiments
- AI can recommend your restaurant. Can it tell a guest what you actually serve tonight?
AI can already find an open table. It cannot fix a restaurant whose hours, menu and details disagree from one listing to the next.
Field notes
- Your front desk knows the answer. AI agents need it too.
AI assistants have started booking beauty appointments. The calendar is the easy part. The front desk's knowledge is the hard part.
Field notes
- Your eyewear catalog was built for optical software. AI agents are next
Eyewear frame data is unusually well standardized. A shopper's AI assistant will ask what it was never built to answer: is this frame, in this color and size, at this store, today?
Field notes
- AI agents are coming for B2B product discovery
The first personal AI assistants book dinner and run errands. The question for a building-products company is what happens when buyers bring them to work.
Field notes
- What happens when an AI agent requests the quote?
A quote request is the safest first thing to let an AI assistant do in building products. It has value, it has bounds, and a person still makes every commitment.
Systems
- AI agents have a part-number problem
A technician reads a part number off a failed bearing. The next five minutes decide which distributor gets the order. Software is about to spend more of them.
Field notes
- Vehicle inventory was only the beginning
We built a car dealership's inventory for AI assistants first. Then we looked at RVs, boats, tractors and excavators, and almost every assumption broke.
Field notes
- AI can find the part. Can it show the part fits?
Every parts counter hears: 2021 F-150, front pads and rotors, I need them today. AI assistants are starting to hear it too.
Field notes
- Technical product data was never designed for AI agents
Electrical, HVAC and plumbing distribution built some of the best product data in wholesale, for catalogs, estimating and EDI. A contractor's AI assistant asks something else: the right part, at a branch, today.
Field notes
- An AI assistant emailed a car dealer. The interesting part is what it asked for.
One dealership inquiry is not a market study. But what it asked for shows what a business needs once customers hand tasks to software.
Field notes
- We read OpenAI's plugin rules before building one. One idea died on the first page.
We read OpenAI's plugin rules before building one. The directory is real, and so is the review. One idea ended on the first page.
Research
- Start with the business, not the homepage
A redesign that starts with the business, not the homepage, changes the structure of the site before any page is designed.
Field notes
- Tools that score how AI mentions a business have a trust problem, and it isn't a secret
Buyers of tools that score how AI mentions a business keep saying the same thing: nobody trusts the numbers.
Research
- The most interesting AI work is happening around the model
Seven AI tools and patterns we examined in one week, each given a single verdict.
Research
- What if the expensive AI is doing the wrong job?
We are testing whether a small AI model can handle the many small judgments around big-model work, without being allowed to act.
Experiments
- Is the cheapest model actually cheaper?
A cheaper AI model cost less per step and more per finished job.
Experiments
- Sales Desk
Our client sites run on checked facts, so our sales record does too. One system carries a customer from first conversation into delivery.
Systems
- How Lehvel's AI systems hand work to each other
We got tired of carrying prompts and status between AI systems by hand. So we built a shared system that does it, with GitHub as the record.
Systems
- Running operations by voice
Owners do not update forms. We are testing whether an owner can say what changed and have it become a change to the business record.
Experiments
- The AI Observatory
We would rather show no number than relabel ordinary search data as an AI score.
Research
- How owners actually manage their digital presence
Five fixed questions, then let the owner talk.
Research
- What browser-level QA found that green tests missed
A passing build is not a passing check.
Field notes
Learn
Dealer and channel data
- Dealer directory data is wrong in more places than one
Manufacturer locators, distributor portals and aggregator feeds each publish a different copy of a dealer's data. Where the copies diverge and who can correct them.
- Dealer network data: who decides which fact is right
Dealer network data governance assigns each fact an owner and an order of precedence across manufacturer, distributor and dealer, before an assistant settles it.
- Duplicate business entities across dealer networks
Duplicate business entities appear when one location exists as several records across manufacturer, distributor and map systems. Resolve them to one record.
- How AI assistants choose between manufacturer, distributor and dealer sources
A product question draws on a manufacturer, a distributor and a dealer, each authoritative for different facts. How assistants weigh them, and where it breaks.
- What AI assistants ask car dealers, and where the answers live
What AI assistants ask car dealers about one vehicle: trim, colors, VIN, arrival and lease. Why it isn't on your website, and why “unknown” beats a stale “no.”
What AI says and why
- Trace a wrong AI answer to its source
A wrong answer from an AI assistant has a source: a directory, a data file, a PDF or a dealer page. How to find it, classify it and decide who corrects it.
- Measuring AI answer accuracy against your own records
Accuracy is measured against a record of what is true, with probes that test identifiers and measurements. A method for scoring each AI answer about a company.
- Who can change a business's published facts
The public, platforms, aggregators and partners can change a company's published facts. Who can overwrite what, how a change is detected and why the source matters.
- AI connectors for manufacturer catalogs: who is authoritative
Assistants are gaining ways to query a manufacturer's catalog directly. The app-store lesson, and the question every connector forces: who is authoritative.
- What AI agents need from a supplier
A purchasing agent needs a supplier's catalog, specifications, availability and authority stated as data. What it reads, and what it cannot do without.
- Controlling the company facts you share with AI
Which company facts may be given to an AI system, under what terms, and with what record. Classification, plan terms, retention and audit for business data.
- OAI-SearchBot robots.txt: the setting that controls ChatGPT Search
OpenAI documents an OAI-SearchBot robots.txt rule for ChatGPT Search. It is separate from the GPTBot training crawler, and Claude uses its own.
Product data
- Part numbers and measurements that disagree across systems
One item carries a part number, a distributor SKU, a GTIN and an ERP code, and a size in nominal, actual and metric forms. Where they diverge and what governs them.
- The product information chain: from PIM to page to feed
The product information chain is the path a fact travels from PIM to page to feed. Where each hop transforms it, and how to find the hop that introduced an error.
- What makes a B2B catalog ready for AI agents
An agent-ready catalog states each fact once, in agreement everywhere, with a unit, a source, a date and an owner. The criteria and how to test them.
- Structured data that shows where a fact came from
Product and Organization markup can carry identifiers, units and dates so software can read where a fact came from. Structured data shows the source, and has limits.
- Does ChatGPT read PDF spec sheets? The PDF problem
Does ChatGPT read PDF spec sheets? It can, but a value buried in an undated PDF is hard to find and trust. Put key facts on the page and link the PDF.
- Can AI understand your product specifications?
AI search for building product manufacturers depends on specs stated clearly: the configuration, the value, the test standard and a current approval.
- What is Product Truth?
What is Product Truth? A company's own record of what it sells, where each fact came from and when it was confirmed, so sites and AI assistants agree.
- Product Truth vs a PIM
Product Truth vs PIM: a PIM manages and sends product content. Product Truth answers questions about products, with the source and date attached.
