For 25 years, procurement software has asked suppliers to change. Register for a portal. Learn a new interface. Upload your documents. They refused — not because they’re unsophisticated, but because email works fine, and they sell to dozens of buyers who each have their own portal.
I spent years on the other side of that wall. A $200M snack brand runs $80–100M in ingredient spend through a shared inbox, a couple of Google Sheets, and the institutional memory of whoever’s still on the team. Quotes as PDFs. CoAs with inconsistent naming. Spec revisions sent as reply-all. None of it structured. All of it the actual system of record.
The first time I watched an LLM read a supplier’s PDF quote and spit out structured prices, MOQs, lead times, and expiration dates — with no portal, no login, no behavior change asked of the supplier — I understood what had shifted. It wasn’t the demand. It was the capability.
“Supplier portals won’t die because they failed technically — they’ll die because AI removes the very reason they existed.” — Bob Solomon, former SVP, Ariba
Every procurement leader I talk to has a version of the same story. A certificate expired, nobody caught it, production stopped. A spec lived in one inbox and missed the co-man. A team paid $4.50 per pound for an ingredient available at $1.50 because they hadn’t run a competitive bid in two years.
These aren’t edge cases. They’re the operating rhythm. And the fix was never another portal.
“In food and beverage, the biggest margin leak rarely comes from price — it comes from constant coordination just to get complete, accurate information.” — Lee Kornfeld, former CPO, Nestlé Health Science and The Bountiful Company
The mistake that defined the last generation of procurement tools was assuming humans must change. Machines can adapt instead. Capture the supplier email that was already coming. Extract the quote, the spec, the cert. Let the data compound.
Email was always the system of record. We finally have something that can read it.
