AI & Supply Chain
ISM World 2026: CPOs Tell Procurement Teams to Slow Down on AI Spend
At ISM World 2026 in Colorado, senior procurement executives — including the CPO of American Airlines and the VP of procurement at The Mosaic Company — told attendees that managing AI costs means starting with low-risk pilots and scaling incrementally. The message was direct: define the business problem first, prove the value, then expand.
Agentic AI Is Targeting the Long Tail — 1,000s of Low-Value POs Nobody Was Managing
A growing body of analysis points to agentic AI's most compelling use case: the enormous volume of low-value transactions that procurement teams have never had bandwidth to properly manage. AI agents handling tail spend negotiation and PO processing autonomously are delivering measurable savings from spend that was previously invisible.
Tail spend has always been the dirty secret of procurement — we know it's costing us, but we don't have the bodies to do anything about it. If agentic AI genuinely closes this gap, the ROI case practically writes itself. The critical caveat: your data quality needs to be clean enough for the agent to act on it. Garbage in, autonomous garbage out.
Supply Chain Tariff Volatility Is Creating a New AI Use Case: Quote Validity Monitoring
Ongoing tariff uncertainty and commodity fluctuations — particularly around copper and critical minerals — are forcing manufacturers to shorten quote validity windows significantly. Procurement teams are now managing a faster-moving pricing environment than at any point in the past decade.
Shorter validity windows compress your decision cycle. That's where AI actually earns its keep right now — not in replacing strategic judgment, but in reducing the time you spend on the groundwork before you can apply that judgment. Running rapid supplier comparisons and market checks in minutes instead of hours may be the most immediately defensible AI investment in the current environment.
DEEP INSIGHT
The 4% Problem: Why AI Pilots Stall Before They Scale
Those three statistics tell the same story from three different angles. Almost every senior procurement professional is personally using AI tools. Nearly half of procurement teams ran a pilot. But almost none of them scaled it.
The gap isn't a technology problem. It's an organisational one.
Here is what I've observed - and what the research consistently confirms. When AI pilots stall, it's almost never because the tool didn't work. It's because the pilot was designed to test the technology rather than to solve a specific business problem. The proof of concept ran in a controlled environment, disconnected from the messy reality of actual workflows, fragmented data systems, and procurement teams that were already at capacity.
"AI adoption in procurement is an organisational challenge, not a software upgrade. Successful AI initiatives require aligned people, processes, and platforms — not isolated pilots layered onto legacy workflows." — Supply Chain Management Review, Feb 2026
I want to give you the practitioner version of this. The organisations that have actually scaled AI in procurement share one characteristic: they identified a single, specific, high-friction task - not a transformation agenda- and built confidence from there. Supplier risk briefings. First-draft RFQ documents. Spend category analysis. One workflow, properly done, becomes the proof of concept for the next one.
The organisations that stalled did something different. They launched a "procurement AI initiative." They bought a platform. They ran a workshop. They got consultants in. And then they discovered that the platform's recommendations were only as good as their master data - which nobody had cleaned in three years. The pilot results were inconclusive. The budget ran out. Back to spreadsheets.
The 4% who scaled weren't necessarily smarter or better-resourced. They were more specific. That specificity is available to every procurement team regardless of organisation size.
TOOL SPOTLIGHT
Perplexity Pro — AI Research · Supplier Intelligence · Market Analysis
✓ Recommended · ~$20/month · ⚠ Verify Sources
I've been using Perplexity Pro this week specifically for supplier research and market briefings. Here's what I tested and what I found.
