The main generative AI use cases in procurement, in rough order from easiest to hardest to put into production, are: drafting RFQs and RFPs from a specification; classifying spend and surfacing tail and maverick purchases; reviewing contracts and extracting clauses; normalizing supplier quotes that arrive in different formats into a like-for-like comparison; flagging line-item outliers against history or should-cost; sourcing long-tail and MRO items that were never worth manual effort; supplier discovery and risk summaries; negotiation support; and intake assistants that answer policy and status questions. The common thread is reading and reconciling unstructured text.
Why This Matters
Adoption is broad and shallow. In 2025 McKinsey found about 40% of procurement functions had begun using generative AI, and in 2026 the Hackett Group found only 12% running AI at large scale, with 69% accessing it through capabilities embedded in platforms they already use rather than a separate product. The gap is between trying a use case and depending on one, and the order above is a guide to which cases survive contact with real supplier data.
How It Works
| Use case | What the model reads | What it produces | Difficulty |
|---|---|---|---|
| RFQ and RFP drafting | A specification or part list | A structured request | Low |
| Spend classification | Line items across systems | Categories, tail and maverick spend | Low to medium |
| Contract review | Agreements | Clauses, obligations, renewals | Medium |
| Quote normalization | Quotes in PDF, email and spreadsheet form | A comparable table | Medium to high |
| Line-item and should-cost checks | Quote lines against history | Flagged outliers | High |
| Tail and MRO sourcing | Long parts lists | Requests routed to suppliers | Medium |
| Supplier discovery and risk | Web and filings | Candidate lists, summaries | Medium |
| Negotiation support | Quote history, terms | Briefs and counter-proposals | High |
| Intake assistants | Policy and system data | Answers to requester questions | Low |
Deloitte's 2025 CPO survey put spend analytics (53.4%), RFP and RFQ generation (42.3%) and contract summarization (41.3%) as the top generative AI investment priorities, all generative reading-and-drafting tasks. Quote normalization and line-item checks are the two least covered in published material and the two where buyers lose the most hours, because that is where sourcing stalls. Each use case, with named examples and the evidence for each, is in generative AI in procurement: nine real use cases. Where generative work shades into agents that act on it is covered in what to let an AI agent do in procurement.
FAQ
Which use case should a team start with?
The highest-frequency, lowest-risk one for that team. For most buyers that is RFQ drafting or quote normalization: drafts are reviewed before anything is sent, and a wrong extraction is caught at comparison rather than at a supplier.
Is generative AI the same as procurement automation?
No. Rules-based automation routes structured steps, such as approvals; generative AI reads and drafts unstructured language, such as a quote in a PDF or a contract clause. They work together, and most value today comes from the generative layer handling the text the rules-based layer could not.
Why do most generative AI pilots stall?
Scope. Teams pilot a broad "AI for procurement" ambition instead of one frequent task with a measured baseline. The Hackett Group's 12%-at-scale figure reflects pilots that could not show what improved.
People also search for: