Why healthcare collaboration fails: Leaders point to incentives, not technology

In healthcare, the standard approach to solving a difficult problem is more collaboration. Health systems are told the future will be co-created, that progress depends on partnership, that everyone has to come together. The executive who opened the CHIME 2026 Innovation Summit in Fairfax, VA is not convinced. 

Matt Kull, chief information and digital officer at Inova Health System, which hosted the summit, argued that healthcare does not have a too-little-collaboration problem. Its committees and advisory boards already meet constantly. What it has, he explained, is an authenticity problem, built on conflicting incentives, siloed rewards, and a shortage of shared accountability. Much of what gets called partnership is really just buying: a vendor sells a finished product, and the meetings around the deal do not make it something they built together.

Kull grounded the claim in a pilot graveyard familiar to most CIOs. When a promising tool fails after launch, teams blame the technology. The more common culprit, he said, is that a clinician showed up on a Tuesday morning and found it built for someone else’s workflow. Every failed pilot in his career traced to the same root: the people measured for on-time delivery and those measured for patient outcomes chased different goals with the same work.

The stakes run higher than any single pilot, Kull argued, because the technology no longer sets anyone apart. Software that once took months can be built in minutes, and no health system holds an edge on tools for long. The durable advantage is how well it works with everyone who has a stake: clinicians, engineers, patients, security, finance. We can all build, he said. What separates organizations is who they choose to build with.

Healthcare silos form around incentives and rewards, not org charts

Silos are not structural, Kull argued. They form around incentive and reward programs, and survive even inside a single IT department, where security, data, and infrastructure answer to different measures. Breaking them down means sharing risk, truth, and upside, not redrawing an org chart.

A later panel on collaborative ecosystems tested the same idea. Asked to name the one thing that most threatens collaboration, one leader said silos, or teams that talk collaboration in a meeting, then do what they had already planned. Another named incentives and money, warning that without aligned goals, collaboration becomes coordination with a better brand. Kull named fear, admitting that giving up control of Inova’s kiosk program to an operations leader on his own team felt terrifying. Inova’s counter is shared performance targets across the management team, so accountability does not split by department.

Co-creation cut Inova’s measles contact tracing from 60 hours to six minutes

Inova offered its clearest evidence that the approach pays off. Needing more infusion capacity, leaders asked clinicians and infusion techs to describe a friction-free workflow, then anchored the redesign with AI. The same rooms, chairs, and staff went from 160 to 210 infusions a day with less exhaustion, and the new demand filled within two days, opening access for patients who had waited years.

The infusion result reflected a principle several speakers returned to: when people help shape a tool from the start, it becomes theirs and they use it.

A measles case offered an even stronger example. Contact tracing that once took seven minutes per record and 60 hours to clear 600 patients now runs at 1.7 seconds per record and finishes in six minutes, sorted by vaccination risk. Quality, clinical, nursing, and contact-tracing teams built it together, and Epic is folding the work into its product for other customers.

In his keynote, Inova’s president of clinical services, Dr. John Moynihan, described the culture behind those numbers. The system had run as a federation of five hospitals with separate medical staffs, and its unified strategy required a shift from “my patient” to “our patient.” More than 400 clinicians and staff met in intentionally inefficient 16-week cohorts to rebuild care around evidence. Asked afterward what the best part was, the biggest word in the response cloud was “collaboration.”

Moynihan was explicit about the mechanism beneath the cohorts. Teams that interact regularly build what he called social capital, the trust that makes collaboration work. Inova organized clinicians into geographic pods, each led by a primary care physician and a specialty partner who share patients and data, and held town halls where physicians who had only traded referrals met in person.

Patients are already co-creating care with AI, with or without their providers

Patients already co-create their care without their clinicians, Kull noted, plugging lab values into ChatGPT before a doctor can respond. An Amazon Web Services employee who has lived with type 1 diabetes for nearly 40 years asked why his providers question him about alcohol, home safety, and his personal life but never about the AI tools he uses to manage his condition.

A fishbowl on co-creating with patients, one of several open-chair panels, supplied cautionary history. Zafar Chaudry, former CIO of Seattle Children’s, recalled ordering 277 registration kiosks at roughly $10,000 each during an Epic rollout, over his own objection, because operations insisted. Not one was used, and staff turned them into coat racks. No one had asked patients what they wanted.

The same instinct had paid off for Chaudry elsewhere. His hospital ran a monthly patient and family advisory group, parents and caregivers alongside patients as young as twelve, and it changed what the team built. When a young patient said the cable television was terrible, his team built a Minecraft version of the hospital, which proved genuinely therapeutic for children facing treatment. His kiosks failed because no one asked. The Minecraft build worked because someone did.

A Valley Health leader pressed the point that collaboration is human before it is technical. In an earlier role as a medical director, his team lifted patient satisfaction scores with no technology at all, just a dress code, chairs set so clinicians sit at eye level with patients, and a required handshake at the start of each visit. Scores rose from the 2nd percentile to the 65th in six months, and to the 85th within a year. At the moment of life and death, he said, technology is not what saves you.

AI can undermine collaboration by shutting down debate

The most pointed exchange asked whether AI helps teams work together at all. Kull argued it can do the opposite by becoming a cudgel for being right. He described colleagues who end a debate by announcing that ChatGPT says the other person is wrong, which closes curiosity instead of opening it.

Tyler Martin of MedStar Health offered a fix borrowed from software engineering. His team routes AI output through the equivalent of a pull request, where several people comment, question the assumptions baked into it, and send it back to a human before it moves ahead.

For all the ways AI can divide people, several speakers said its governance had done the opposite. Deciding how a health system wants to use AI, and be known for it, created a forum that had not existed before, pulling clinical, technical, and operational leaders into one conversation. Lauren Ross, who leads AI governance at Mayo Clinic, said the goal is to meet teams where they are, so that transparency levels the field rather than slowing people down.

Health systems can share on technology and compete on outcomes

Stephanie Hines of Valleywise Health rejected the premise that collaboration requires someone to lose power. The wrong word is “lose,” she argued. The right one is “share.” Competition, inside a system and across it, grows from a scarcity mindset, the belief that resources are fixed and must be guarded. Once teams treat those resources as collective and give everyone a voice, whoever holds the most has reason to direct it toward the shared goal.

Kull drew the line between competing and sharing directly. Inova competes on outcomes and experience, not technology, he said, and would share anything technical with the MedStar, Hopkins, Mayo, and Valley leaders present, since that is not where they compete. Chappy Van Katessen, Inova’s chief quality and safety officer, cited learning health networks as proof. In one pediatric collaborative focused on inflammatory bowel disease, remission rates climbed from 55 percent to more than 80 percent over a decade through shared data and network-wide goals, an improvement the member hospitals agreed none could have reached alone.