Supplier portals failed because they required suppliers to change behavior. AI kills the portal by extracting structured data from the email suppliers already send — no logins, no uploads, no “supplier enablement” campaigns.
For two decades, procurement software has chased the same dream: if suppliers would just log into our system, upload their documents, and fill out our forms, we’d finally have structured data. The dream kept not happening. Procurement kept paying the consultants. And suppliers kept sending email.
The portal was a rational idea for a pre-AI world
Before large language models, the only way to get structured data at scale was to force the structure at the point of entry. Make the supplier pick from a dropdown. Make them upload a file with your fields. Make them fill in your schema before you’ll talk to them.
That’s what supplier portals did. And in a pre-AI world, it was the only option.
Suppliers rationally refused
The problem: suppliers don’t have one customer. They have dozens or hundreds. Every buyer wanted them to learn a different portal, remember a different password, upload a slightly different certificate format.
A cinnamon vendor supplying 37 bakeries was never going to maintain 37 different buyer portals. Email is the universal adapter — the one system every stakeholder already uses. So suppliers quoted via email, sent CoAs via email, responded to RFPs via email, and treated the portal as a compliance chore they did under duress.
”Supplier enablement” was a tell
An entire consulting sub-industry emerged around convincing suppliers to use portals — training campaigns, reminder sequences, financial incentives. Large procurement consultancies built real businesses on this work.
When software requires a specialized consulting practice to get adoption, the software is the problem. Supplier “enablement” was a euphemism for “our system fights the way this market actually works, and we’ll pay you to muscle it in.”
Bob Solomon, a former SVP at Ariba, has put it bluntly: suppliers never wanted another password or another upload portal. Any system that ignores that reality is already obsolete.
AI inverts the assumption
Large language models extract structure from unstructured inputs. That single capability flips the default.
You no longer need suppliers to type their quote into your fields. You can let them email the quote in whatever format they prefer — a PDF, an Excel attachment, a paragraph of prose — and extract the price, MOQ, spec, lead time, and certification details programmatically. Same for CoAs, spec sheets, production schedules, and incident reports.
The portal’s core job — getting structured data from suppliers — can now happen silently, on your side of the wire.
Why CPG feels this first
Food, beverage, supplement, and pet food procurement is one of the most document-heavy categories in the economy. Every ingredient carries specs, certifications, allergen declarations, and batch-specific CoAs. Much of that data lives in PDFs and email threads.
CPG procurement spent the last twenty years trying to fit that reality into portal-shaped software. It didn’t fit. The shift to AI-native extraction is biggest here precisely because the mismatch was biggest here.
The new model: intelligence where work already happens
Instead of pulling suppliers into your workflow, you layer intelligence onto theirs. Email stays the interface. Structured data emerges automatically. The supplier file — every quote, every certification, every conversation — becomes a living network that strengthens with each interaction instead of a static database someone has to maintain.
That’s the model Waystation is built on. No supplier logins. No portal. No “enablement.”
