U.S. Medicare CMS launches new AI-native payment rules: Use AI Agent to reduce costs to make money

Core conclusion

The U.S. Centers for Medicare & Medicaid Services (CMS) officially launched the ACCESS (Advancing Chronic Care with Effective, Scalable Solutions) program in April 2026 - a 10-year AI-native medical payment pilot. This is the first time that the U.S. government has given the green light to AI-driven medical services in the health insurance system: not paying for services, but paying for health outcomes.

Key Points

• Date of occurrence: Announcement of the first batch of selected institutions on April 30, 2026
• Coverage: 150 million Medicare/Medicaid beneficiaries, first 150 institutions selected
• Core changes: Medical insurance shifts from payment based on projects to payment based on health outcomes, and AI becomes a key capability
• The first batch of selected examples: the voice AI Agent "Flora" deployed by the AI nursing company Pair Team has processed thousands of patient interactions

Background: Intergenerational changes in medical insurance payment models

CMS is the largest single payer in the United States, covering approximately 150 million people with health insurance. In the past few decades, its payment model has always been based on one principle: Pay for Service (Fee-for-Service). How many examinations the hospital has done and how many consultations it has completed will be settled on a project-by-project basis.

This model provides little incentive for doctors or institutions to improve patients' long-term health. More importantly, it does not leave room for AI - there is no mechanism to pay for "an AI Agent to proactively call the patient between consultations to follow up on the condition, coordinate housing referrals, and remind medication."

That's what the ACCESS program is going to change.

Key Figure: According to Pair Team research, 1 in 4 hospitalizations will never occur and 1 in 2 emergency room visits will be avoided under its community-based integrated care model. AI Agent eliminates the need for large-scale manual teams.

CORE DESIGN OF THE ACCESS PROGRAM

ACCESS was designed by two heads of CMS Innovation Center with VC backgrounds – Abe Sutton (former Rubicon Founders investor) and Jacob Shiff (Chief AI and Technology Officer, CMS Innovation Center). Their design ideas bear the distinct imprint of a startup company:

A counter-intuitive design: CMS’ per-patient-month payments were lower than many participants expected. However, the CMS Innovation Center believes that this is precisely the design intention - The payment rate is low enough so that participants have the incentive to use AI to reduce costs to the extreme.

Real case: Pair Team's AI Agent "Flora

Pair Team was one of the first agencies selected for ACCESS, founded in 2019, with a special focus on vulnerable patients who face a combination of chronic illness and unstable housing/food insecurity. The company currently has approximately 850 clinical professionals, operates the largest community health workforce in California, and generates annual revenue in excess of nine figures.

Nine months ago, Pair Team deployed voice AI Agent Flora as the primary patient interface. Flora is available 24 hours a day to handle initial screening of patient consultations, coordinate referrals, and provide proactive follow-up between clinical consultations.

Founder Batlivala recounts an example that transformed his perception: a 67-year-old woman living in her car while managing PTSD and congestive heart failure. "Flora may have been the only person she could communicate with about her condition for weeks."

Pair Team's study (peer-reviewed in the Journal of General Internal Medicine) shows that this model significantly reduces avoidable emergency room and hospitalization utilization.

Opportunities for AI practitioners

The ACCESS program sends a bigger signal than it appears:

1. The compliance industry is paving the way for AI – Healthcare is one of the most highly regulated industries. Institutions like CMS not only do not ban AI, but actively design incentive mechanisms. This is an important reference for AI Agents to enter the compliance industry.

2. Pay-per-result AI agent model - If your AI tool can directly improve a measurable result (increased conversion rate, lower refund rate, shortened processing time), rather than providing abstract efficiency improvements, it will be more likely to be adopted.

3. Voice AI is becoming infrastructure - Flora proves that voice AI Agent has a huge space to replace manual labor in patient management (essentially user operations) scenarios. The same logic can be applied to other industries such as customer service, insurance claims, property management, etc.

4. Digital health financing picks up - Digital health financing in Q1 of 2026 hit the highest level since the epidemic, with AI companies accounting for the largest share. ACCESS’s nominees range from AI diagnosis and treatment startups to wearable device companies like Whoop, covering a wide range of topics.

Extended thinking

ACCESS is not without its pitfalls. Critics noted:

• Participants are inputting extremely sensitive patient data (housing status, mental illness, poverty levels) into federal infrastructure
• The CMS Innovation Center’s previous record has been less than stellar — a 2023 Congressional Budget Office analysis shows the Center instead increased federal spending by $5.4 billion in its first decade
• Data privacy challenges are significant, especially for the vulnerable populations ACCESS is designed to serve

But it is undeniable that this time the government did not "adapt" to AI, but took the initiative to design an economic engine for AI.

Related reading

• AI Agent-Driven Content Automation: n8n MCP Building Guide from Scratch
• He Built an AI Automation Stack with Claude + n8n — $4K to $12K/mo in 6 Months