hunterAI

Part 4: Why Continuous Audit Becomes a Strategic Necessity For Procurement & Supply Chain Last week, Starbucks retired NomadGo, its AI-powered inventory system, across all 11,000+ North American stores, nine months after launch. After claiming 99% accuracy and speed over manual counting. Datature's 2026 Enterprise Vision AI Adoption Report identifies "distribution shift", the gap between training data and production conditions, as the second most common cause of computer vision deployment failure. In other words: the data the AI was trained on didn't reflect the reality it was asked to operate in. The problem wasn't the technology itself, but the operational foundation. It is the pattern behind most AI implementations that disappoint: AI does not eliminate structural weaknesses in underlying data and operational logic. It amplifies them. But data readiness is not a one-time event The deeper lesson from Starbucks isn't just about launching on clean data. It's that even a validated foundation drifts. Shelf conditions shift. Products move. Environments change. A system accurate at deployment becomes unreliable over time and without a mechanism to detect that drift, the gap between what AI sees and what reality looks like grows silently. This is precisely the structural problem cybersecurity identified a decade ago. Controls that tested correctly at implementation decayed between audits and acting on stale assurance was catastrophic when it materialized. The discipline's response was to formalize continuous validation as a named category: "Continuous controls monitoring (CCM) is a set of technologies to reduce business losses through continuous monitoring and reducing the cost of audits through continuous auditing of the controls in financial and other transactional applications." — #Gartner IT Glossary Procurement is approximately ten years behind that conversation. Gartner's Market Guide for Healthcare Provider Supply Chain Data and Analytics Solutions puts a precise number on it: annually, 30% to 40% of items in a typical item master become obsolete or replaced. A health system's foundational data layer doesn't just drift, it turns over by a third every single year, across item masters containing up to 150,000 records, with hundreds of thousands of items under contract. A vendor base that looks clean at the moment of an AI implementation will look materially different twelve months later. Without continuous audit, that degradation is invisible until it's already embedded in every downstream decision the system makes. The category that healthcare procurement still hasn't named Healthcare procurement needs to designate vendor master continuous audit as a formal category and establish it as the first step and universal standard for every AI initiative. Not a cleanup project or a one-time assessment before go-live. A standing control, so that implementation happens on data that's actually ready, and stays that way.

The AI healthcare market is projected to grow from $36.7B to more than $505B within the next decade. At the same time, healthcare supply chain technology investments continue accelerating across procurement, finance, operations, and revenue cycle transformation. Behind this explosive growth sits a much more difficult question: How are healthcare leaders actually expected to make rational technology investment decisions in a market evolving faster than procurement cycles themselves? New episode of The Health Intelligence #Pitch is now live! https://lnkd.in/ewFurEec This time we explore how supply chain and procurement executives navigate billion-dollar technology decisions in an environment shaped by analyst reports, peer influence, consultants, AI-generated research, and rapidly shifting market narratives.

Part 3: Why Cost Optimisation Fails Without Vendor Layer Integrity For supply chain and finance leaders cost optimisation has been a strategic priority for years. And challenges to meet the targets can ground into data rather than strategy itself. The cost of bad data is not hypothetical Hospital expense growth is outpacing revenue: supply chain expenses are up 26% and drug costs up 31% since 2022, while Kaufman Hall's adjusted operating margin closed out 2025 at just 1.3%. In that environment, every savings initiative matters. But those initiatives are only as precise as the data they run on. Poor supply chain data integrity, including item master issues, contributes to an estimated $25.4 billion in annual overspend across U.S. hospitals. That figure partially explains the persistent gap between negotiated savings and realised savings that supply chain leaders report year after year. The progression nobody maps The vendor master data problem produces a cascade of problems, each layer of error building on the one before. 1️⃣ Attributes break first. Manufacturer names truncated, catalog numbers mismatched, GTINs missing. These gaps block product comparison, contract validation, and accurate reimbursement. 2️⃣ Duplicates multiply on top of that. The same product appears multiple times with slightly different descriptions or IDs. Spend splits across records, standardisation opportunities disappear, and analysts spend hours reconciling conflicts that shouldn't exist. 3️⃣ Velocity compounds both. Suppliers make roughly 10 million item data changes every year, and GPOs process about 30,000 contract changes every month. A system that isn't continuously tracking those changes accumulates drift on top of already degraded data. What changes when AI enters the picture Gartner's AI Maturity Benchmarks for Healthcare Providers (2026) identifies data as the primary constraint on AI ROI in healthcare: organisations are actively moving AI from pilot into production, with the expectation that analytics outputs will inform decisions at speed and at scale. When AI models run on supply chain and spend data, they don't apply judgment, they apply pattern recognition to whatever the data contains. In the Market Guide for Healthcare Provider Supply Chain Data and Analytics Solutions confirms that the item master is the primary source of truth for supply chain transactions, stored across fragmented ERP, inventory, and EHR systems. That fragmentation was always a problem. Salil Joshi Now it's a problem that AI amplifies rather than solves. "Without disciplined oversight and standardisation, incremental cost creep can quickly erode financial performance while offering little clinical value." Which itself opens a new discussion while Health systems are investing millions in AI implementation those investments drain rather than deliver. And the incremental value to patient care remains near zero. Not because AI failed, but because the foundation was never ready.

New Episode of #Pitch Is Live! https://lnkd.in/eE5fCiWy The final episode in our series dedicated to analyzing the GPO model, its transformation potential, hidden inefficiencies, and ultimately its impact on patient care. In this final discussion, we raise some of the most uncomfortable questions in healthcare supply chain and finance, questions that still deserve attention and reflection: Who actually benefits from the lack of transparency? As healthcare costs continue to rise, the conversation goes far beyond pricing. We examine the structural realities behind GPO models, distributor incentives, contract compliance, supplier dynamics, and the “last mile” accountability gap that no single party truly owns. Our guests discuss: 👉 Why transparency remains difficult across the healthcare ecosystem 👉 Whether current incentive structures discourage full accountability 👉 How legacy operating models impact innovation and patient outcomes 👉 Why measuring true value remains one of healthcare’s biggest challenges Featuring candid perspectives from Bill Kopitke and Jeremy Strong on the future of healthcare supply chain transparency, innovation, and system-wide accountability.

Part 2: For Supply Chain and Finance Leaders: The Invisible Layer Undermining Spend Integrity When we begin working with a health system, finance leadership almost always says the same thing: their vendor data is "generally clean." The data tells a different story. The vendor master underlies every invoice approval, every payment run, every fraud detection flag — maintained through periodic manual reviews, reactive cleanups, and institutional trust that the data is mostly right. That combination is quietly costing you. 1️⃣ The scale of the problem is hiding in plain sight Gartner's Market Guide for Healthcare Provider Supply Chain Data and Analytics Solutions (May 2025) reports that annually 30% to 40% of items in a typical item master become obsolete or replaced. A third of your records go stale every year by default. Across the 78 health systems we have analyzed, vendor lists shrink by up to 30% following a structured audit — duplicates, ghost vendors, and inactive entities carried for years without anyone flagging them. 2️⃣ Vendor spread turns a data problem into a financial control problem Gartner (January 2026): "Supplier proliferation is an urgent challenge. Left unchecked, this complexity creates cost leakage, dirty and unstructured data and undermines resilience." The report poses a question every finance leader should be asking: "Is your supplier portfolio a strategic asset or a hidden liability?" And yet most organizations still treat vendor master management as a project, something that happens and is considered done. 3️⃣ The operational drag compounds the exposure As vendor portfolios proliferate, the administrative burden grows with them, quietly consuming sourcing capacity and diluting accountability. In one of our client engagements, the team was consuming approximately 86 hours per month on labor-intensive AP anomaly management before automation. That's not a control. That's a structural inefficiency and with humans in the loop on a process this manual, error rates rise with volume. 4️⃣ Fraud detection is only as good as the data underneath Gartner defines fraud detection as the "ability to flag duplicate payments and invoices, errors, and other anomalous transactions" (Critical Capabilities for AP Applications, March 2025).  But every one of those signals depends entirely on the vendor master below.  The questions worth sitting with Have you ever quantified, not estimated, quantified the financial losses tied to vendor data quality issues or fraudulent activity? How many FTE hours per month does your team spend validating and reconciling supplier information? And what is that time not being spent on? If you don't have a continuous audit process for your vendor master how confident are you, really, in the integrity of the data driving every financial decision downstream? The value is there. It's buried under a process that was never designed to find it.

Part 2: For Supply Chain and Finance Leaders: The Invisible Layer Undermining Spend Integrity When we begin working with a health system, finance leadership almost always says the same thing: their vendor data is "generally clean." The data tells a different story. The vendor master underlies every invoice approval, every payment run, every fraud detection flag — maintained through periodic manual reviews, reactive cleanups, and institutional trust that the data is mostly right. That combination is quietly costing you. 1️⃣ The scale of the problem is hiding in plain sight Gartner's Market Guide for Healthcare Provider Supply Chain Data and Analytics Solutions (May 2025) reports that annually 30% to 40% of items in a typical item master become obsolete or replaced. A third of your records go stale every year by default. Across the 78 health systems we have analyzed, vendor lists shrink by up to 30% following a structured audit — duplicates, ghost vendors, and inactive entities carried for years without anyone flagging them. 2️⃣ Vendor spread turns a data problem into a financial control problem Gartner (January 2026): "Supplier proliferation is an urgent challenge. Left unchecked, this complexity creates cost leakage, dirty and unstructured data and undermines resilience." The report poses a question every finance leader should be asking: "Is your supplier portfolio a strategic asset or a hidden liability?" And yet most organizations still treat vendor master management as a project, something that happens and is considered done. 3️⃣ The operational drag compounds the exposure As vendor portfolios proliferate, the administrative burden grows with them, quietly consuming sourcing capacity and diluting accountability. In one of our client engagements, the team was consuming approximately 86 hours per month on labor-intensive AP anomaly management before automation. That's not a control. That's a structural inefficiency and with humans in the loop on a process this manual, error rates rise with volume. 4️⃣ Fraud detection is only as good as the data underneath Gartner defines fraud detection as the "ability to flag duplicate payments and invoices, errors, and other anomalous transactions" (Critical Capabilities for AP Applications, March 2025).  But every one of those signals depends entirely on the vendor master below.  The questions worth sitting with Have you ever quantified, not estimated, quantified the financial losses tied to vendor data quality issues or fraudulent activity? How many FTE hours per month does your team spend validating and reconciling supplier information? And what is that time not being spent on? If you don't have a continuous audit process for your vendor master how confident are you, really, in the integrity of the data driving every financial decision downstream? The value is there. It's buried under a process that was never designed to find it.

For Supply Chain & Finance leaders: What Cybersecurity Just Revealed About Invisible Risk The Mythos case made headlines for the right reasons. Anthropic's model identified thousands of high-severity vulnerabilities across major operating systems, vulnerabilities that existed undetected because they were never systematically measured. The question is: is it actually relevant only for cybersecurity, or does this case reveal a structural problem that we can encounter across different disciplines? A closer look at healthcare procurement seems to be a perfect example of a system that contains many vulnerabilities due to its design. Healthcare supply chains operate across a fragmented network, GPOs, health systems, suppliers, manufacturers, where data moves constantly, but without shared structure or consistent governance. The same product gets classified differently across systems. Vendor data is duplicated or misaligned. Information is incomplete, inconsistent, and hard to reconcile. Yet this same data drives demand forecasts, contract negotiations, and cost optimization. It is time to ask: can you actually rely on top-level data and performance or optimization outcomes if your underlying data is essentially a black box?

Unlike more centralized environments, procurement in healthcare operates across a fragmented network of stakeholders, where data continuously moves between GPOs, health systems, suppliers, and manufacturers without a shared structure or consistent governance. This fragmentation exists at multiple levels. The same product can be attributed in different ways, creating a landscape where information is inconsistent, incomplete, and difficult to reconcile, with direct implications for financial accuracy. The issue is pressing because data sits at the center of operational and financial decision-making. But beyond that, the real question is: what is actually the scale of potential losses or other vulnerabilities caused by the lack of data transparency? 🎙️ In the latest episode, we explore the opportunities within the rapidly evolving technology landscape that can help solve data challenges for health systems. The question we discuss in depth with our guests Bill Kopitke Jeremy Strong is following: Is it possible for evolving AI platforms to become a new layer above the GPO ecosystem to provide visibility? 💬 Watch the full episode to learn more about top experts’ perspectives https://lnkd.in/dHbnvTRg

The speed of change should make us all pause. In 1995, the Internet was a phenomenon. Everyone saw that this was the milestone that would change everything. Ten years down the line, Amazon, online banking, food delivery, these services were shaping the reality of daily life. No one was talking about the Internet anymore. It had become invisible infrastructure, the foundation everything else was built on. That path took appriximately 10 years. The parallel with LLMs is striking. In 2022, Foundation Models were at the Peak of Inflated Expectations on Gartner's Hype Cycle for Emerging Technologies. ChatGPT launched on November 30, 2022, reached 1 million users in 5 days, and 100 million within two months. It was the fastest adoption in software history. It was a phenomenon. By 2026, the term "Foundation Models" has largely disappeared from the narratives, replaced by "LLMs." But more importantly, it is already clear that LLMs are too generic and not efficient enough for real enterprise problems. At the consumer level, AI-powered applications for content, video, and custom queries are launching daily. At the industrial level, the shift is moving toward domain-specific language models and creating a real FOMO effect. If you are still heavily investing in integrating a generic model and trying to justify that investment, the window may already be closing. Gartner predicts that by 2027, more than 50% of enterprise models will be domain-specific, up from just 1% in 2023. It took the Internet 10 years to move from phenomenon to invisible foundation. AI did it in 4. The question is: how do you keep up with that pace, especially at the enterprise level, where significant investments are already on the table? And what reasoning do you bring to the board when everything is shifting this fast? The answer is vendors with short time-to-value (rear tangible value) and contingency-based pricing.

Purchased services represent up to 45% of non-labor operating costs and yet it has no SKU, no commodity code, and no unit of measure. IT, legal, facilities, staffing. Everything that keeps the organization running but doesn't fit neatly into the supply chain taxonomy that was built for clinical goods. The irony is that clinical spend is actually the easier problem. You can track a surgical glove, because the system has language for it. Purchased services, on the other hand, is what one of our guests described, very accurately, as "a little squidgy." There's no standard way to benchmark it. There's no GPO contract structure that travels cleanly from negotiation to invoice. And for years, the working model has essentially been: get your arms around the spend first, understand what you're actually dealing with, and then, well, maybe, figure out how to drive value from it. What makes this particularly interesting is the structural reason why. GPOs built their entire value proposition around the clinical side — volume aggregation, contract leverage, rebate flows. That model works when you can attach a commodity code to the thing being purchased. The moment you move into services, that logic starts to break down. Not because GPOs aren't trying, but because the data architecture was never designed for it. And health systems, feeding incomplete item masters upstream, don't make it easier. The real question, the one we spent time on in this episode, is who actually owns the accountability gap. And the answer is everyone and no one at the same time. We explored this in depth in the latest Episode of Healthcare Intelligence Pitch podcast with two people who have lived on both sides of this equation: Bill Kopitke, who spent years inside the GPO and technology world, and Jeremy Strong, Chief Supply Chain Officer at Rush University Medical Center, who deals with the operational consequences of this ambiguity every day. 🎙️ Check out the latest episode: https://lnkd.in/dr4fjam2