We've been asking the wrong question.
"Will AI replace procurement jobs?" generates heat but little light. The better question is: which specific tasks in which specific roles are genuinely ready for AI — and which aren't, and why?
To answer it properly, we decomposed five Sourcing & Contracting roles into their constituent tasks — every trigger, input, action, and output — and scored each one against a simple framework: is AI ready to do this autonomously today, partially with human oversight, or not ready at all?
The result is the AI Readiness Benchmark for Procurement, live now at air.procurement.news.
What we found
AI performs best on structured, data-intensive tasks with clear inputs and outputs: spend cube analysis, RFx document drafting, contract data extraction, compliance tracking, reporting. These tasks have something in common — they don't require you to read the room.
AI hits its ceiling on tasks that require judgment under ambiguity, relationship navigation, and accountability. A contract manager deciding whether to exercise a termination right needs to weigh legal exposure, commercial relationships, organizational risk appetite, and timing. No model is carrying that.
We identified six recurring reasons AI falls short in procurement work: context limitations, judgment requirements, relationship management, tacit knowledge, legal accountability, and creativity. The distribution across roles tells you a lot about where the real automation opportunity lies — and where it doesn't.
Critically, not all tasks are equal. Automating a low-effort administrative task moves the needle less than automating a high-effort analytical one. Our scoring weights each task by actual human effort, so the role-level scores reflect what matters: how much of your time AI can genuinely displace, not just how many boxes it can tick.
What's in the dashboard
The interactive benchmark at air.procurement.news lets you:
See weighted AI readiness scores for each of the 5 roles
Drill into individual tasks with full AI ceiling reasoning
Filter by task type, effort level, and readiness status
Run a scenario analyzer: describe a real procurement situation and get a personalized AI readiness assessment
Why this matters now
Procurement teams are being asked to evaluate AI tools with no independent reference point. Vendors benchmark their own products. Analysts produce high-level frameworks. Nobody has done the task-level decomposition that actually tells you whether to buy, build, or wait.
This benchmark is a starting point. We've covered Sourcing & Contracting first — five roles, 58 tasks. If the research resonates, we'll expand to all 35 procurement roles.
Explore the benchmark → air.procurement.news
Research by procurement.news and allcaps.ai . Methodology available on the dashboard.