Latest / The Buyer's Side with Fexingo: Procurement, Sourcing, and B2B Purchasing Decisions / How Procurement Teams Are Using AI to Spot Supplier Risk
Transcript
- Lucas: So there's this mid-sized manufacturer based in Ohio — makes industrial valves for HVAC systems — and they have about 2,200 suppliers spread across 40 countries. Luna: That is a lot of moving parts for a company most people have never heard of. Lucas: Exactly. And until two years ago, they were using spreadsheets and a couple of junior analysts to track supplier health. Then they deployed an AI platform called Resilinc, and within the first month it flagged something they had completely missed. Luna: What was it? Lucas: A microchip supplier in Taiwan — it was their sole source for a specific controller chip — and the AI had picked up on a pattern of labor unrest and a pending regulatory change in that region. The platform assigned it a high-risk score. Luna: And this was before any major disruption? Lucas: About four months before. The company used that window to qualify a second supplier in Vietnam and pre-order an extra six-month inventory buffer. When the Taiwan supplier eventually had a three-month shutdown due to a strike, they barely felt it. Luna: So the AI basically bought them time. That's the kind of story that makes procurement sound almost glamorous. Lucas: Which is not the usual vibe of the category. But seriously, supplier risk is one of those areas where artificial intelligence is moving from nice to have to table stakes. A McKinsey survey from late 2025 found that firms using ai driven risk monitoring reduced supply disruptions by 32 percent year-over-year. Luna: That's a huge number. What kind of risks are we talking about beyond labor strikes? Lucas: Financial instability is a big one. The AI can scrape public filings, news, credit reports — even social media sentiment — to detect if a supplier is heading toward bankruptcy. Weather events, geopolitical tension, port congestion. It's essentially a giant pattern-recognition engine running on thousands of data streams. Luna: And this is different from the old approach of just checking D&B reports once a quarter? Lucas: Night and day. The old approach is reactive — you find out about a problem when your shipment doesn't arrive. The AI is predictive. It gives you a probabilistic score for each supplier, updated in near real time. The Ohio manufacturer told me their risk detection window went from about two weeks to over ninety days. Luna: Ninety days to act on a warning. That changes the negotiation dynamic too, doesn't it? You're not scrambling, you're planning. Lucas: Absolutely. And that's where the procurement team starts to look less like a cost center and more like a strategic function. But I want to be careful not to oversell it. The technology has real limitations. Luna: Such as? Lucas: False positives are a problem. The same McKinsey survey noted that about 15 percent of ai generated risk alerts turned out to be noise. If you're a procurement manager getting pinged three times a week with false alarms, you start ignoring the system. Luna: So you need human judgment to filter that. Which means the AI doesn't replace the analyst, it just changes what the analyst does. Lucas: Exactly. The best implementations I've seen pair the AI with a senior procurement professional who knows the supplier relationships personally. The AI says 'look at this supplier,' the human says 'I know that supplier — here's the context.' Together they make better decisions. Luna: What about cost? A lot of our listeners are at companies with maybe a hundred suppliers, not two thousand. Is this only for the big players? Lucas: It's getting more accessible. Resilinc and similar platforms like Prewave and SourceDay now offer tiered pricing based on supplier count. You can start with a hundred-supplier subscription for maybe thirty thousand dollars a year. That's not nothing, but compare it to the cost of a single unplanned disruption. Luna: What's the typical cost of a disruption? I've seen numbers all over the map. Lucas: It varies hugely by industry, but a 2024 study from the Institute for Supply Management estimated that the average manufacturing firm loses $184 million per year due to supply disruptions. For a mid-market company, a single event can wipe out six months of profit. Luna: So thirty thousand a year for an AI tool starts to look like cheap insurance. Lucas: That's exactly how the Ohio manufacturer framed it. They said the AI paid for itself in the first quarter by preventing one near-miss with a secondary supplier that had been showing signs of distress. The platform flagged a shipping company that was about to have its license revoked — dodged a potential two-week delay on a critical component. Luna: I want to circle back to the human element. You mentioned the analyst needs to have context. What does that look like in practice? Lucas: Well, the AI can tell you that a supplier's employee sentiment on Glassdoor has dropped sharply, or that their CEO just sold a large block of stock. But it can't tell you that the supplier's plant manager just had a heart attack and operations are shaky. That kind of information comes from relationship building. Luna: So the AI handles the signal, the human handles the story. Lucas: Right. And the best procurement teams are training their people to interpret the AI outputs — to ask 'why is this score elevated?' and then pick up the phone. The technology is a force multiplier, not a replacement. Luna: Are there any sectors where this is particularly advanced? Automotive? Pharma? Lucas: Pharma is probably the leader because of regulatory pressure. If a drug ingredient supplier goes down, you can't just switch — you need FDA approval. So they have massive incentives to predict disruptions. Automotive is close behind, especially after the chip shortage in 2021 and 2022. Luna: The chip shortage really was a wake-up call for the entire procurement profession. Lucas: It was a generational event. It made boards of directors pay attention to what had been a back-office function. And that attention has fueled investment in tools like AI risk monitoring. I think we're going to see this become standard practice within five years. Luna: What about smaller suppliers? The ones that don't have big digital footprints for the AI to scan? Lucas: That's a real gap. If a supplier is a small family-run business with no web presence, no news coverage, no credit history — the AI has very little to work with. Some platforms are trying to fill that gap by pulling data from logistics partners and payment platforms. But it's still a blind spot. Luna: So the human relationship becomes even more critical there. You actually have to visit the factory. Lucas: Exactly. No shortcut for that. The AI is only as good as the data it consumes, and small suppliers often fly under the radar. Luna: Let me ask a slightly provocative question. If AI gets good enough at predicting risk, does procurement become a purely analytical function? Do we lose the art of negotiation? Lucas: I think the negotiation part actually becomes more strategic. If you know a supplier is at risk, you can negotiate different terms — shorter contracts, more frequent audits, alternative sourcing clauses. The AI gives you leverage because you know something the supplier may not know you know. Luna: That's a fascinating dynamic. It changes the information asymmetry in the buyer's favor. Lucas: Exactly. But of course that only works if you act on the intelligence. I've seen companies invest in the tool but then not change their behavior. They still do annual supplier reviews. The AI is screaming at them, and they ignore it. Luna: Classic implementation failure. The tool is only as good as the process around it. Lucas: Right. And that's where leadership matters. The companies that succeed are the ones that redesign their procurement workflows around the AI, not just bolt it on top of existing processes. Luna: Give me an example of a redesigned workflow. Lucas: Sure. One company I spoke with — a medical device maker — now has a daily morning huddle where the procurement team reviews the top five AI alerts from the last 24 hours. They assign each one to a category owner who has to respond within 48 hours. That's a huge shift from the old monthly review meeting. Luna: So the cadence changes from monthly to daily. That's a cultural shift. Lucas: It is. And it requires trust in the system. But once the team sees that the AI catches things they would have missed, that trust builds quickly. Luna: I'm wondering about the ethics of this. Is there a risk of over-surveillance? Of treating suppliers as high-risk based on incomplete data? Lucas: That's a legitimate concern. If the AI flags a supplier in a developing country based on political instability, does that bias procurement decisions against that region? You could end up with a kind of algorithmic redlining. Responsible buyers are aware of this and build in guardrails. Luna: What kind of guardrails? Lucas: Some companies require a human override for any decision to drop a supplier based solely on AI risk scores. Others use the AI to trigger deeper investigation rather than automatic action. You want to avoid a situation where the algorithm kills a relationship that could have been managed. Luna: So the best practice is ai assisted, not ai automated. Lucas: Exactly. Especially in procurement, where relationships matter. You don't want to lose a good supplier because of a false flag. Luna: What do you see as the next frontier for AI in procurement? Beyond risk monitoring. Lucas: I think we're moving toward AI that can help with supplier negotiation — suggesting optimal price points and contract terms based on market data and the supplier's risk profile. Some companies are already piloting systems that simulate negotiations. But that's still early stage. Luna: Simulated negotiations? That sounds like something out of a sci-fi movie. Lucas: It does. But the logic is straightforward: the AI can process thousands of past negotiations, market conditions, and supplier behaviors to recommend a strategy. It's like a chess engine for procurement. The human still makes the final move, but the computer helps you see the board better. Luna: I can see that being really powerful, especially for companies that handle thousands of contracts a year. Lucas: Exactly. And it could level the playing field for smaller procurement teams that don't have deep category expertise. The AI becomes a kind of institutional memory. Luna: Alright, but let's bring it back to the Ohio valve manufacturer. What's the one thing they would tell another company just starting with AI risk monitoring? Lucas: They told me: start with your most critical suppliers, not all of them. Get the AI tuned on the top 50 where you have the most data, prove the value, then expand. Don't try to boil the ocean. Luna: That's solid advice. And it applies to most technology implementations. Lucas: It does. And I think that's where we are with AI in procurement right now — the tools are proven, but the adoption curve is still early. The companies that figure out how to integrate them thoughtfully are going to have a real advantage. Luna: So for our listeners in procurement: start small, focus on the critical few, and don't ignore the human element. Lucas: That's the takeaway. And maybe pick up the phone and call a supplier you haven't talked to in a while. The AI will help you know when to call, but only you can build the relationship. Luna: Good note to end on. Thanks Lucas. Lucas: Thanks Luna. Talk to you next time.