Most conversations about artificial intelligence in healthcare drift toward the clinic. Models that read scans, flag patients at risk of sepsis, draft discharge notes.
Those advances are real and they deserve the attention they get. Yet much of what decides whether a patient is treated on schedule happens far from the exam room, in the long chain of orders, shipments, stock counts and reconciliations that moves a medical product from a distributor’s warehouse to the tray beside an operating table.
When that chain breaks, patients feel it. A September 2026 report from the VA Office of Inspector General found that the main operating room at the VA Augusta Health Care System in Georgia closed four times between October 2025 and February 2026 because of incomplete supply kits, expired stock and unreliable restocking. One closure lasted 27 days.
The health system had already awarded a $3.2 million contract for an automated inventory platform, but it was never fully installed, a former supply chief disabled it in 2023, and inspectors found some of its equipment abandoned in a warehouse.
Augusta had bought the technology; It never became part of how the hospital actually ran.
Follow a single medical supply from start to finish and the trouble becomes clear. It passes through distributors, pharmacies, clinics and laboratories, each running on its own systems, spreadsheets, email threads and phone calls.
The information needed to keep shelves stocked usually exists somewhere, but it is scattered, and people end up stitching it together by hand. That work eats time, invites mistakes and drives up costs in a sector where precision is supposed to be non-negotiable.
A study in the Annals of Internal Medicine put administrative costs at 34.2% of total U.S. health spending in 2017, and consulting firm Chartis warned in July 2026 that running supply chains on margins that are too thin can hurt product availability, disrupt clinical workflows and affect patient outcomes.
Better algorithms alone won’t fix that. The harder task is getting dozens of disconnected systems to behave like one.
Keeping people in charge of the decisions that matter
Adrián Figueroa has spent years close to that problem. An electrical engineer and graduate of the Stanford Graduate School of Business, he founded Flexio, a tech-enabled logistics services company specializing in healthcare.
Routina grew out of that experience. Launched in 2024, it uses software and AI to help medical product distributors and other healthcare organizations connect information across their systems and cut down on manual coordination.
For Figueroa, adding AI is the easy part. The challenge is getting the data and processes of every player in the chain to work in sync, with automation giving teams more information and capacity rather than taking them out of the loop.
That raises the question of oversight, since a mistake in healthcare operations has consequences. Figueroa doesn’t believe a human needs to check every output. “With AI as it stands now, the volume of output is more than a human can process,” he said.
“Reviewing everything turns the human into the bottleneck, but human judgment is still needed for high-risk or low-confidence decisions.”
His approach is to have the AI assign a confidence score and a risk score to every result. People review and approve only what comes back as low confidence or high risk, while the system tracks how well it performs on the outputs it treats as safe. That review has to mean something.
“Human oversight has to be real, not a person clicking ‘approve’ all day. Specific people own those flagged decisions, they have the authority to override the system, and every override is logged and reviewed.”
Regulators are moving in the same direction. Article 14 of the EU AI Act requires high-risk systems to allow effective human oversight, including the ability to override them, and similar debates are gaining ground across the Americas.
Figueroa also stresses transparency about what is measured and why. “If the input data is biased, AI will apply that bias faster and at scale,” he warned. And on accountability he was blunt. “What AI shouldn’t do is make the final decision.”
A first test in the Dominican Republic
Those ideas are now being put to work. Surgical Solutions International, a surgical services and medical products distributor, recently launched a new line of medical products in the Dominican Republic with Cardinal Health as part of a strategic alliance.
“We will be implementing Routina’s technology at Surgical Solutions to manage all of Cardinal Health’s products and those of future suppliers,” Figueroa noted.
Routina also plans to pitch itself directly to Cardinal Health “as a solution to manage all of its distributors in the different countries where it operates.”
If that happens, the test moves from a single distributor in Santo Domingo to a multinational network, where a missing kit in one country can look a lot like the one that kept an operating room in Augusta dark for 27 days.
Featured image courtesy of Routina
Disclosure: This article mentions a client of an Espacio portfolio company.
23 c
18 c
14 c