Summary: CMS Innovation Center’s ACCESS model is a 10-year Medicare program starting July 5, 2026 with 150 organizations, including Pair Team. It pays a monthly per-member fee supplemented by bonuses tied to measurable chronic-care outcomes, giving AI-driven care a reimbursement pathway. The article flags privacy risk and notes a 2023 CBO analysis found the Innovation Center’s first decade raised federal spending by $5.4 billion.
One of this year’s weightier AI stories is sliding past nearly the whole American tech world as of May 2026: Medicare’s innovation arm is assembling a payment model designed expressly for AI. Known as ACCESS — Advancing Chronic Care with Effective, Scalable Solutions — the 10-year CMS Innovation Center program has 150 chosen organizations starting on July 5, 2026, paid for the health outcomes they deliver, not activities performed. Among them is Pair Team, a chronic-care startup told of its acceptance on April 30.
Why does it matter? Payment has always been the choke point for AI in healthcare: however capable the model, with no code to bill under it stays outside the care pathway — and ACCESS constructs that mechanism for the first time.
How ACCESS Works
The design: pay per member, topped up by bonuses tied to outcomes. Organizations collect a steady monthly fee per member, in full only once patients clear measurable targets — blood pressure moving down, pain easing. The scope: chronic disease — diabetes, hypertension, chronic kidney disease, obesity, depression, and anxiety.
The model’s architects make an odd pairing for CMS: the Innovation Center’s director, Abe Sutton, came out of venture investing, and chief AI and technology officer Jacob Shiff previously founded a healthcare company. The opening 150 spans connected-device firms, virtual nutrition providers, startups building AI doctors, and wearable companies — Whoop included.
What Kept AI Out of the Traditional Reimbursement System
Classic Medicare pays for clinician hours. An AI agent watching patients between appointments, arranging housing referrals, or checking that a prescription got collected is left without any billing line to claim. In TechCrunch’s words, ACCESS “creates that mechanism for the first time.”
Pair Team CEO Neil Batlivala names it “a payment model transformation”: “You just couldn’t do this before.” That also fits the direction clinical AI evidence keeps pointing in — value emerging from patients’ continued engagement with a system, as emergency-room AI diagnosis research shows (/blog/harvard-ai-er-diagnosis-study/).
Flora at Pair Team: The Case Already in Motion
Founded in 2019, Pair Team works with chronic-condition patients who also face food insecurity, housing instability, or transportation gaps — about a third of all Americans. About 850 clinical professionals are on staff; it calls its community health workforce California’s biggest, reports nine-figure revenue, and around $30 million raised from Kleiner Perkins, Next Ventures, and Kraft Ventures. A peer-reviewed Journal of General Internal Medicine study found its care trims preventable ER and inpatient use; by Batlivala’s telling, a quarter of hospital visits and half of ER visits never happen at all for its patients. Its partnerships put some 500,000 potential patients in reach, and the stated goal is one million within three years.
Nine months back, the company rolled out its voice-based AI agent Flora, which takes check-ins, referrals, and intake nonstop. Its standout case: a 67-year-old woman whose home is her car, living with PTSD and congestive heart failure, who stayed on the line with Flora for over an hour. Batlivala called it “both incredible and depressing” — “That’s the companionship piece,” and “it turns out that is truly an intervention.”
The Low Rates Are Intentional, Not a Bug
Per-patient monthly fees landed beneath what many participants had bargained for, and Batlivala reads that as deliberate: “the reimbursement rates have to be low,” since “the economics only work if you’re running a lean, AI-first operation.” Put differently, ACCESS props up no labor-heavy delivery model — it makes participants anchor their cost structure in AI.
Two Risks to Watch: Data Privacy and the Innovation Center’s History
Two risks deserve a hard look. Privacy: sensitive patient information will end up inside federal systems that carry a breach history — providers’ Social Security numbers leaked out of a CMS directory before. Money: per a 2023 CBO analysis, the Innovation Center’s first ten years left federal spending $5.4 billion higher. “Savings” is a hypothesis still awaiting proof, not a promise.
Behind it all, capital is flowing back into the sector: no digital health funding quarter since the pandemic has topped Q1 2026, and AI supplied most of the momentum. Once the payment side opens, this cohort will outrun any oversight — and ACCESS’s 10-year clock sets the experiment’s time limit.