How to Use This Glossary
Procurement AI has accumulated a vocabulary that mixes old procurement terms, finance jargon and machine-learning language. This glossary defines the 50 terms that matter most in 2026, each in one or two precise sentences you can quote directly. It is organised into four groups — process, technology, metrics, and governance — so you can find a term by where it lives in the discipline.
For deeper treatments, several terms link to a full guide. If you want the strategic context behind the vocabulary, the State of Procurement AI 2026 report frames how these concepts fit together, and the CPO strategic guide translates them into priorities. You can also browse the site-wide glossary hub.
Quick orientation: if you read only five entries, read agentic procurement, source-to-pay, spend under management, human-in-the-loop, and touchless invoice. They anchor most procurement AI conversations.
Process & Lifecycle Terms
- 1. Source-to-pay (S2P)
- The end-to-end procurement lifecycle from sourcing and supplier selection through contracting, ordering, receipt and payment. See source-to-pay vs procure-to-pay.
- 2. Procure-to-pay (P2P)
- The downstream subset of S2P covering requisition, purchase order, goods receipt, invoice and payment.
- 3. Strategic sourcing
- The structured process of analysing spend, running competitive events and selecting suppliers to optimise total value, not just price.
- 4. Intake-to-procure
- The front-door process that captures a purchase request and orchestrates it through approvals, sourcing and ordering. See the intake-to-procure category.
- 5. Guided buying
- A buying experience that steers employees toward preferred suppliers and compliant catalogues at the point of need.
- 6. Category management
- Managing a group of related spend (a category) as a strategic portfolio with its own plan. See the category management framework guide.
- 7. Tail spend
- The large number of low-value, fragmented transactions that fall outside managed categories. See what is tail spend.
- 8. Three-way matching
- The control that compares purchase order, goods receipt and invoice before payment. See the three-way matching reference.
- 9. Contract lifecycle management (CLM)
- The management of contracts from request and drafting through negotiation, execution, obligations and renewal.
- 10. RFx
- The umbrella term for structured supplier requests: RFI (information), RFP (proposal) and RFQ (quote).
- 11. e-Auction
- A live, competitive online bidding event in which suppliers submit improving bids against each other.
- 12. Supplier onboarding
- The process of qualifying, validating and registering a new supplier into procurement and finance systems.
- 13. Maverick spend
- Purchasing made outside approved channels or contracts, undermining negotiated savings and compliance.
- 14. Direct spend
- Spend on goods and materials that go into a finished product. See direct vs indirect procurement.
- 15. Indirect spend
- Spend on goods and services that support operations but are not part of the product, such as IT, travel and facilities.
- 16. Should-cost modelling
- Estimating what a product should cost to produce, from materials and labour up, to inform negotiation.
Technology & AI Terms
- 17. Agentic procurement
- Procurement in which AI agents autonomously execute multi-step tasks within human-set guardrails. See the definitive guide.
- 18. AI agent
- Software that perceives context, plans, and takes actions toward a goal with limited human direction.
- 19. Procurement copilot
- An assistive AI layer embedded in a procurement tool that answers questions and drafts work, with a human acting. See the copilots category.
- 20. Generative AI
- Models that produce new content — text, summaries, drafts — used in procurement for RFP drafting, contract summarisation and supplier comms.
- 21. Large language model (LLM)
- A model trained on vast text that underpins most procurement copilots and document AI.
- 22. Natural language processing (NLP)
- The branch of AI that interprets human language, used to read contracts, invoices and supplier messages.
- 23. Optical character recognition (OCR)
- Technology that converts document images into machine-readable text; the capture step in invoice automation.
- 24. Spend classification
- Automatically mapping each transaction to a spend category. See our classification accuracy benchmark.
- 25. UNSPSC
- The United Nations Standard Products and Services Code, a hierarchical taxonomy of segments, families, classes and commodities.
- 26. Autonomous negotiation
- AI that conducts a negotiation with a supplier — proposing and counter-proposing — within set limits. See the negotiation AI category.
- 27. Retrieval-augmented generation (RAG)
- A technique that grounds an AI's answers in retrieved source documents to reduce errors.
- 28. Grounding
- Tying an AI's output to verified data sources — your contracts, catalogues or spend — so answers are traceable.
- 29. Hallucination
- A confident but false AI output; a core reason procurement keeps a human in the loop on high-value decisions.
- 30. Straight-through processing
- Completing a transaction end to end with no manual touch; in AP, the touchless invoice path.
- 31. Supplier enrichment
- Augmenting supplier records with external data — what they sell, risk signals, certifications — to improve matching and analytics.
- 32. API-first integration
- An architecture in which a tool exposes and consumes data through APIs, easing connection to ERPs and other systems.
Browse the tools behind the terms
Every concept here maps to real products. Compare them side by side or explore by category.
Metrics & Value Terms
- 33. Spend under management
- The share of addressable spend actively sourced and controlled by procurement. See the definitive guide.
- 34. Addressable spend
- The portion of total spend that procurement can realistically influence or control.
- 35. Touchless invoice rate
- The share of invoices processed with no human intervention. See our touchless processing data.
- 36. Cost avoidance
- Savings that prevent a future cost increase rather than reducing the current price paid.
- 37. Cost savings
- A measurable reduction in price paid versus a defined baseline.
- 38. Cycle time
- The elapsed time to complete a process, such as requisition-to-PO or sourcing-event duration.
- 39. Time to value
- How long after go-live a tool delivers measurable benefit.
- 40. Total cost of ownership (TCO)
- The full multi-year cost of a tool: licence, implementation, integration, data and change. See the pricing & TCO index.
- 41. Match rate
- The share of invoices successfully matched to a PO and receipt without exception.
- 42. Classification confidence
- A score indicating how certain the AI is about a category assignment, used to decide what auto-classifies.
Governance, Risk & Compliance Terms
- 43. Human-in-the-loop
- An automation design where AI handles routine cases and escalates uncertain ones to a person.
- 44. Autonomy level
- A scale describing how independently a tool acts, from assistive to fully autonomous.
- 45. Audit trail
- A complete, time-stamped record of who or what took each action, essential for procurement controls.
- 46. Explainability
- The ability to show why an AI reached a recommendation, increasingly a procurement-policy requirement.
- 47. SOC 2
- An independent attestation that a vendor meets defined security, availability and confidentiality controls.
- 48. Scope 3 emissions
- Indirect supply-chain emissions, the procurement-owned majority of most companies' carbon footprint. See Scope 3 explained.
- 49. Supplier risk scoring
- A model-driven score of a supplier's financial, operational, ESG and compliance risk.
- 50. Vendor lock-in
- Dependence on a vendor that makes switching costly, a key risk to weigh when standardising on a suite.
Where to Go Next
A glossary is a map, not the territory. Once the vocabulary is clear, the useful next step is to see how the terms behave in real decisions: read the CPO strategic guide for the leadership view, the best tail-spend AI shortlist for a worked category, and the category management framework for the discipline that ties spend strategy together. To assess specific products against these concepts, start at the comparison hub or browse the blog.
Frequently Asked Questions
What is agentic procurement? Procurement in which AI agents autonomously carry out multi-step tasks within human-set guardrails, rather than only assisting a person. It sits above copilots and rules-based automation on the autonomy curve.
What's the difference between source-to-pay and procure-to-pay? S2P covers the full lifecycle including upstream sourcing and contracting; P2P is the downstream requisition-to-payment subset.
What does UNSPSC mean? The United Nations Standard Products and Services Code, a hierarchical spend taxonomy of segments, families, classes and commodities used by spend analytics tools.
What is a touchless invoice rate? The share of invoices processed end to end with no human touch — the cleanest measure of AP automation effectiveness, best read as a blended figure.
What is human-in-the-loop AI? A design where AI handles high-confidence routine cases automatically and escalates uncertain or high-value decisions to a person. It is the dominant procurement AI operating model in 2026.