Pallets of machined metal parts on a machine shop floorDistributor · Deep parts research

The problem

Customers know what has broken but not what the replacement is called. One writes “transistor” and means a switch. Another writes “seal” and means a gasket. Catalogue search matches words, so these enquiries return nothing.

The salesperson then investigates by hand: guess the part family, open drawings one by one, search manufacturer sites, then email a supplier. Each ambiguous enquiry took minutes to hours of research, and some went unanswered while a competitor replied.

How it works

An AI research agent handles these enquiries automatically. It processes each quote request, searches the company’s parts library and manufacturer websites, works out the likely part and returns a part number with the pages and links it used. It places the order with the supplier or manufacturer the customer specifies.

  • Plain-language enquiry in, part number and evidence out
  • Searches the internal library first, then manufacturer sites
  • Every answer links to the drawing page or web page it came from
  • Follow-up questions stay in the same research thread
Parts research
Customer says the “transistor” on their fryer has gone. Model FE-40, bought 2019.
Searched parts library · 3 drawings for FE-40
Checked manufacturer parts list
Compared candidate parts against the description
The part is most likely the high-limit safety thermostat, part 3471-006. Customers often call it a transistor because of its shape and wiring. It sits behind the control panel on the 2018–2021 FE-40.12
1 Drawing FE-40, control panel, p.6Internal library
2 Manufacturer parts list, FE seriesWeb
Add to quoteAsk a follow-upAnswered in 21 s
Concept: The research thread. The agent shows the steps it took, the likely part and the sources behind the answer. Sample data.

The results

+6ppquote conversion on enquiries that catalogue search misses
Minutesto resolve an enquiry that took hours of research
  • Ambiguous enquiries get a checked answer the same hour instead of waiting for a supplier.
  • The salesperson sees the drawing page or web page behind every answer, so no part number is quoted unchecked.
  • Research that depended on experienced staff can now be done by anyone on the desk.
  • Fewer supplier requests, so fewer quotes stall for days.