AI visibility for materials & manufacturing
Get specified by AI.
Engineers now ask ChatGPT, Claude, Gemini and Perplexity which material to choose, why a part is failing, and which process to use. Reference Grade measures what those engines answer and traces every source they cite. Then we engineer the technical content that makes your products the answer and your pages the reference.
Table view
| Supplier pages | Market research & press wires | Trade press, databases & reference | Total | |
|---|---|---|---|---|
| Claude | 8% (1) | 75% (9) | 17% (2) | 12 |
| ChatGPT | 100% (17) | 0% (0) | 0% (0) | 17 |
The new first step
The spec now starts in a chat window.
Material selection used to begin with a datasheet, a distributor call or a property database. Those steps still happen. Increasingly, though, they come after an AI assistant has already framed the answer: which material class fits, which grades are worth sampling, which suppliers are worth a call. Engineers don't fully trust those answers, so they check them. What matters is whose pages they land on when they do.
of technical buyers use generative AI in the purchasing process.
TREW Marketing & GlobalSpec, State of Marketing to Engineers, 2026
of the buying journey is done before an engineer contacts a vendor.
TREW Marketing & GlobalSpec, State of Marketing to Engineers, 2026
choose the better-known brand when two options are technically equal.
TREW Marketing & GlobalSpec, State of Marketing to Engineers, 2026
From our research
Most of what AI tells engineers about your materials isn't written by you.
We asked ChatGPT and Claude 52 engineering questions across eight segments, from midsole foams to flip-chip underfills, and traced every source behind 156 answers.
Answered from memory
Asked plainly, Claude searched the web for 18 of 52 questions. For troubleshooting and process questions it never searched at all; the answer came from what the model learned in training. Pages engines can't read today are pages future models won't know.
Built from other people's pages
In sourced answers, pages from established material suppliers made up 10% of Claude's citations and 31% of ChatGPT's. Patents, papers, trade press, distributors and smaller suppliers' blogs supplied the rest.
Named, but not sourced
When Claude named an established supplier in a sourced answer, that supplier's own pages backed the mention only 39% of the time. The description of the product came from somewhere else.
Every engine reads differently
For the same question, ChatGPT and Claude shared on average 12% of their cited domains, and on 11 of 52 questions none at all. A strategy tuned to one engine misses the others.
Who gets cited when engineers ask AI
Share of cited URLs by source type
Table view
| Supplier pages | Manufacturers & service providers | Market research & press wires | Patents | Academic papers | Trade press, databases & reference | Other | Total | |
|---|---|---|---|---|---|---|---|---|
| ChatGPT | 44% (145) | 20% (67) | 0% (0) | 0% (1) | 21% (70) | 3% (10) | 11% (38) | 331 |
| Claude, sources requested | 20% (121) | 15% (91) | 3% (17) | 13% (77) | 18% (111) | 12% (73) | 20% (120) | 610 |
| Claude, as asked | 20% (27) | 16% (21) | 8% (11) | 5% (7) | 16% (22) | 11% (15) | 23% (31) | 134 |
Method
Measured like a material. Engineered like a product.
Answer Assay
We build the panel of questions your customers' engineers actually ask, organized by application, role and intent. We run it repeatedly on ChatGPT, Claude, Gemini and Google AI Mode, and parse every answer for the companies, trade names, grades and material classes it recommends.
Source Map and Spec Watch
We trace every citation to the page and passage it came from, and show who wins each question: your pages, a competitor's, distributors, databases, patents, papers or trade press. Spec Watch flags answers that misstate your properties, cite discontinued grades or attribute your brands to someone else.
Content Engineering
A team led by a materials scientist writes and structures the content engines cite: HTML datasheets, selection guides, comparison pages, processing and troubleshooting guides, application notes. Every claim traces to your test data. We publish on your site, or host a Technical Library on your domain.
Crawler Telemetry and re-measurement
We confirm AI crawlers actually fetch the new pages, checking against published crawler IP ranges rather than user-agent strings. Then we re-run the panel and report what changed, with its run-to-run spread.
Specification
Typical properties
| Property | Typical value | Notes |
|---|---|---|
| Engines | ChatGPT, Claude, Gemini, Google AI Mode | Perplexity, Copilot and regional engines on request |
| Question panel | 150–500 questions per business unit | Built from your applications, personas and intents |
| Repeats | 3–5 runs per question, per engine, per cycle | Engines vary run to run; we report rates, not screenshots |
| Cadence | Weekly or monthly; quarterly re-baseline | |
| Parsed entities | Companies · trade names · grades · material classes | With rank and context |
| Citation depth | URL · page type · quoted passage · retrieved vs. cited | |
| Accuracy check | Claims vs. your current TDS and certifications | Spec Watch |
| Crawler verification | AI-crawler fetches checked against published IP ranges | From your server or CDN logs, or your Technical Library |
| Regions & languages | US by default | EU and Asia; German, Japanese, Chinese and Korean on request |
Solutions
Built for the companies engineers specify.
Material & resin suppliers
Polymers, elastomers, foams and fibers. Win the material-class decision, then the grade.
Learn more →Formulation, bonding, surfacesSpecialty chemicals, adhesives & coatings
Be the formulation, bonding and surface answer engineers get first.
Learn more →Packaging, thermal, dielectricElectronic & semiconductor materials
Get into answers currently assembled from papers, patents and market-research teasers.
Learn more →Molders, extruders, machine shops, AM bureausManufacturing service providers
Own the process-selection and cost questions that decide where an RFQ goes.
Learn more →Line cards, equivalents, availabilityDistributors & compounders
Be the source for “where do I get it” and “what's equivalent.”
Learn more →Not sure where you fit?Start with an AI Spec Audit
Tell us what you make and who you sell to. We'll show you what engineers are told.
Request an audit →Outcomes
What changes when your pages are the reference
- Shortlists start with you. Your grades appear when the material class is chosen, not after a competitor has framed the trade-off.
- Your numbers define the comparison. Engines quote your tested values, under your test conditions, instead of an old patent or a distributor's PDF.
- Fewer wrong answers to unwind. Technical service spends less time correcting what an engineer was told by a chatbot.
- Launches show up where engineers look. New grades and sustainability claims can appear in retrieved answers within weeks of publication, and become part of what future models learn.
- Accurate after every acquisition and rename. When brands change owners or grades are renamed, the answers follow.
Who we are
Built by a materials scientist.
Reference Grade was co-founded by Chris Dimitriou, PhD, a materials scientist who spent eight years developing polymers for consumer products at Nike, working alongside the world's leading material suppliers.
The questions engineers now ask AI are materials questions. They deserve answers engineered by people who can read a DMA curve, know what ASTM D395 measures, and won't publish a claim your test data can't support.
Built on Assay
Reference Grade runs on Assay, the AI-answer measurement engine we built and also run for Homiere, our product for real estate. Assay captures AI answers on a recurring schedule and traces their sources. It also verifies crawler activity from both sides: what an engine says it read, and what actually arrived at the server. It is the instrument behind our published research on how answer engines choose their sources.
- Where does a ChatGPT answer come from? Cache, index and live fetch in OpenAI's search stack, measured from both sides.
- When does the model look it up? Parametric knowledge, controlled retrieval and how answer engines arbitrate between them.
Read
Research and notes
Engagements
Start with evidence.
AI Spec Audit
See where you stand in five business days.
Baseline Study
One business unit, measured in depth.
Program
Continuous measurement plus content engineering.
Questions
Frequently asked
How is this different from SEO?
Search rankings and AI answers draw on overlapping but different sources. In our pilot, Claude answered 65% of engineering questions without searching the web at all, and ChatGPT and Claude shared only 12% of cited domains for the same question. AI visibility needs its own measurement.
Which engines do you measure?
ChatGPT, Claude, Gemini and Google AI Mode. Perplexity, Microsoft Copilot and regional engines are available on request.
Can you guarantee our products will be recommended?
No. Engines are probabilistic and change often. We guarantee measurement you can audit and content engineered to the standard engines cite, and we report the change honestly.
Do AI engines read our PDFs and datasheets?
Yes, when they can reach them: 14–20% of the URLs cited in our pilot were PDFs. The bigger problem is delivery. In our audit of 14 flagship grades, only 3 suppliers published a full HTML datasheet; others delivered theirs through JavaScript, third-party catalogs that block crawlers, or firewalls, and a distributor's copy was often the one engines could read.
Should we block AI crawlers?
Separate three kinds of bot. Training crawlers (such as GPTBot and ClaudeBot) feed future models. Search crawlers (OAI-SearchBot, Claude-SearchBot, PerplexityBot) put pages into AI search answers. User-triggered fetchers (ChatGPT-User, Claude-User) read a page when someone asks. Blocking search agents removes you from AI search answers; blocking training crawlers keeps future models from learning your products, and most of Claude's engineering answers in our pilot came from training data. We help you decide deliberately.
Start with evidence
See what AI tells engineers about your materials.
We'll run 25 of your customers' questions across three engines and show you who gets named, which sources get cited, and what's wrong. Free, in five business days.