Blog
Notes from the lab
Short reads from our research on how AI engines answer materials and manufacturing questions, and what suppliers can do about it.
We asked ChatGPT and Claude 52 engineering questions. Here's who they cited.From midsole foams to flip-chip underfills, we traced 1,075 sources behind 156 AI answers to engineering questions. Supplier pages were a minority, patents and papers did much of the talking, and the two engines barely agreed.Read the article →AI answers most engineering questions without looking anything upAsked plainly, Claude searched the web for 18 of 52 engineering questions and never for troubleshooting or process questions. For those, the only content that counts is content the model learned in training.Read the article →AI searches for the brands it already knowsNearly a third of Claude's web searches named a company or trade name before it had read anything. The shortlist comes from memory, and the search confirms it. Incumbency compounds.Read the article →How an AI builds a supplier listAsked which companies supply underfill for advanced packaging, Claude built its list mostly from market-research report teasers and press wires. ChatGPT cited 17 supplier pages. The same question, two very different sources of truth.Read the article →What AI tells a footwear materials engineerBoth engines recommend PEBA foam for maximum rebound, yet the brand most associated with PEBA appeared in none of 24 footwear answers. One supplier's TPU brands appeared in 10. Patents, papers and running blogs did much of the explaining.Read the article →
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.