Last week, market researchers projected the remote patient monitoring market will reach $88.3 billion by 2035, with AI-driven analytics cited as the primary growth engine. The same week, Innovaccer announced it is pouring another $250 million into its agentic AI platform, joining an industry chorus that says intelligent agents will rewrite how care gets delivered. Both stories are real. Both are being read in board rooms across the country as proof that healthcare's AI moment has arrived.
Then there is the data nobody wants to talk about. Recent research from Microsoft and The Health Management Academy found that 47 percent of health system leaders say they are using or assessing AI agents. Only 3 percent have actually deployed them in live clinical workflows. That is the gap that will define the next decade of value-based care, and it is wider than most CEOs and CFOs are willing to admit.
If your organization is betting its workforce strategy, its chronic care P&L, or its 2026 capital plan on agentic AI, you need to understand why the gap exists and what it takes to close it. Otherwise you are buying software that will sit unused while your nurses keep burning out and your readmissions keep climbing.
Demand for technology-enabled chronic care has never been higher. The 2026 CMS Final Rule increased reimbursement for RPM, CCM, APCM, and BHI by roughly 7 to 21 percent, depending on the code. CMS just accepted more than 150 organizations into the ACCESS Model, an outcome-aligned payment pilot for chronic conditions like hypertension, diabetes, depression, and chronic pain. Payers are finally putting real dollars behind the idea that managing chronic disease at home beats treating exacerbations in the hospital.
That is the tailwind. Here is the headwind. The U.S. is short more than 250,000 registered nurses and roughly 85,000 physicians, and the World Health Organization projects a global healthcare worker shortage of 11 million by 2030. There is no version of the next five years where you hire your way out of chronic care management. There is also no version where pure software solves it, because AI without a clinician on the other end of the workflow is a notification engine, not a care plan.
This is why the 3 percent deployment number matters more than the $88 billion market number. The market is pricing in a level of real-world clinical integration that the industry has not figured out how to execute.
Talk to any health system that has tried to deploy agentic AI for chronic care and you will hear the same four problems. Each of them is fixable, but none of them is fixed by buying more software.
The first is accountability. When an AI agent flags a patient with a rising blood pressure trend, who follows up, when, and using what protocol? In most systems, the alert lands in a queue that nobody owns. The second is integration. Most agentic AI tools were built to live inside the EHR or alongside it, not inside the actual workflow of a clinician managing a panel of 200 chronic disease patients. The third is trust. Clinicians who do not understand why an AI flagged a patient will ignore the flag, and they are right to. The fourth is patient adherence. The most sophisticated AI in the world does nothing if the patient does not take the reading, take the pill, or pick up the phone.
These are not algorithm problems. They are operating model problems. Agentic AI will only work in healthcare when the human layer underneath it is engineered with the same rigor as the model layer above it.
The health systems that pull ahead in the next 24 months will be the ones that stop framing AI as a substitute for staffing and start treating it as the multiplier on top of a clinical workforce they actually invest in. That means licensed clinicians paired with AI workflows, not replaced by them.
This is the model we built Welby Health on, and the early data is consistent with what the broader research suggests. Patients in our programs using cellular-enabled blood pressure cuffs see a 20 percent reduction in blood pressure readings. Patients using smart glucose monitors see a 20 percent or greater drop in blood glucose within four weeks. Heart failure patients in our programs are 5.5 times more likely to adhere to their life-saving therapies. None of that happens because of an algorithm alone. It happens because an RN case manager is looking at the AI's prioritized list every morning, calling the right patient at the right time, and intervening before a problem becomes a hospitalization.
The agentic AI does the work no human should be doing. It ingests vital signs, flags adherence gaps, prepares the visit summary, and routes the right patient to the right clinician. The RN does the work no agent can do. They listen, they educate, they negotiate medication reconciliation with a skeptical patient, and they earn the trust that drives a daily blood pressure reading for the next twelve months. Take either layer away and the system collapses.
This is also where the financial math works. CCM, RPM, and TCM codes generate real, billable revenue for partners, and outcome-aligned models like ACCESS reward the same behavior fee-for-service is now finally paying for. But you only capture that revenue if patients stay enrolled, take their readings, and hit measurable health goals. That happens through human relationships, not push notifications.
If you sit in the C-suite of a health system or provider organization, three decisions in front of you will determine whether you are in the 3 percent or the 47 percent at the end of 2026.
The first is honest about what your AI is actually doing today. Not in the demo, not in the pilot, but in the live workflow of your busiest clinicians. If the answer is that it is sitting in an inbox nobody opens, you do not have an AI strategy. You have a software license.
The second is choosing a clinical operating model that scales. You can hire and burn out your way through 2026, or you can pair a smaller, licensed clinical team with AI workflows that do the routine work for them. The math on outsourced or hybrid models is now favorable in a way it was not three years ago, particularly given the reimbursement increases CMS finalized for 2026.
The third is the bet on outcomes versus activities. The ACCESS Model is an early signal of where Medicare is heading. The organizations that build their chronic care program around measurable patient outcomes will win in fee-for-service today and in capitated and risk-bearing arrangements tomorrow. The ones that built their program around activity volume will spend the rest of the decade trying to retrofit it.
The AI hype cycle in healthcare is not the problem. The problem is the gap between the cycle and the clinical floor. The market thinks the gap will close on its own as the technology matures. It will not. It will close only for the organizations that pair real clinicians with real AI, design the human workflow with the same care as the model, and stop pretending that algorithms can hold the hand of a 78-year-old with congestive heart failure at 9 p.m. on a Tuesday.
That is not a software question. It is a leadership question. And the next twelve months will sort the answers.
The remote patient monitoring market is projected to reach $88.3 billion by 2035, but only 3 percent of health systems have deployed agentic AI in live clinical workflows. The gap between AI hype and real-world deployment is where most health systems will lose the next decade. Agentic AI alone does not solve chronic care, because the bottleneck is operating model, not algorithm. The winning model pairs licensed clinicians with AI workflows, supported by Welby's outcomes including a 20 percent blood pressure reduction and 5.5 times higher adherence in heart failure patients. CEOs have twelve months to decide whether they are in the 3 percent that deploys or the 47 percent that assesses.
