Humana's AI Strategy: How the Medicare Advantage Leader Is Using Conversational AI for Senior Care

TL;DR#

Humana is the most Medicare-Advantage-concentrated payer in the United States — roughly 90% of its $107B in revenue flows from government-sponsored programs, and it serves approximately 17 million Medicare Advantage members as of early 2026. That concentration makes Humana the textbook case for why conversational AI insurance interfaces matter: the population it serves is older, more voice-comfortable, and more likely to abandon a 14-field form than any other insurance cohort in the country. After the 2021 sale of Kindred at Home's hospice business and the strategic pivot toward CenterWell primary care, Humana has quietly become an AI-and-care-delivery story masquerading as a health plan. The 2024–2026 Medicare Advantage reimbursement squeeze — a roughly 0.16% effective rate cut for 2025 layered on tighter risk-adjustment rules — has turned every dollar of administrative overhead into a margin lever. Voice-first conversational AI for intake, social determinants of health screening, and post-discharge follow-up is no longer an innovation experiment at Humana. It is the operating model.

Humana at a Glance: The MA-Pure-Play#

Humana is the only top-five US health insurer whose strategy and survival depend almost entirely on Medicare Advantage. Roughly 90% of its premium revenue comes from government-sponsored plans, with around 5.5 million MA members on the books and a footprint of about 17 million Medicare-eligible members served across MA, prescription drug plans, and adjacent care assets. Compare that to UnitedHealth Group, where MA is a large but not existential line, or to Elevance, where commercial and Medicaid balance the book. Humana has no such cushion.

That concentration has two consequences. First, every operational decision at Humana is filtered through "does this work for a 73-year-old with three chronic conditions, possibly hearing-impaired, possibly bilingual, possibly cognitively fatigued?" Second, the company's growth math depends on Star Ratings, risk adjustment accuracy, and member retention — three metrics that all hinge on the quality of conversations the plan can have with members between renewal periods.

The pivot to CenterWell — Humana's payer-owned primary care, home health, and pharmacy business — is the structural answer to that math. CenterWell operates more than 300 senior-focused primary care clinics and a home-health arm built from the unsold portion of Kindred at Home. Humana is no longer just paying claims for seniors. It is delivering the care, capturing the encounter data, and trying to do both with a fraction of the administrative friction of a traditional managed-care operation.

Why Medicare Advantage Is the Hardest Channel for Forms#

Medicare Advantage is the most form-hostile distribution channel in US insurance. The reasons stack:

• Age and cognitive load. The median MA member is 75. Working memory load on a 20-field form is a real comprehension barrier, not a UX nitpick.
• Vision and motor accessibility. Small touch targets, low-contrast dropdowns, and tiny error states on mobile are functional barriers, not preferences.
• Language complexity. Plans must collect data on chronic conditions, social determinants, prescription histories, and care preferences — vocabulary that does not map cleanly to dropdown menus.
• Bilingual reality. A meaningful share of MA members are Spanish-dominant or speak English as a second language. Dual-pathed forms double the cognitive cost.
• Voice familiarity. Older members are demonstrably more comfortable with voice than chat or web forms — this is the only generation that grew up doing business over the phone.

That is the gap a conversational AI insurance interface is designed to close. We've written about why deflection is the wrong goal for conversational AI in insurance — the point is not to keep members away from humans, it's to capture the messy, contextual, "it depends" answers that forms flatten. Nowhere is that more true than at a payer whose members literally cannot complete a 90-page application without help. (See our piece on how conversational underwriting is replacing 90-page life insurance applications for the parallel pattern in life insurance.)

Where Humana Is Investing in AI#

Humana's AI investments cluster around four real, observable workstreams — not the generic "we use AI" disclosures that fill most payer earnings calls.

CenterWell-embedded clinical AI. Inside CenterWell's senior-focused clinics, AI scribes and ambient documentation tools reduce primary-care visit administrative load so physicians can spend more time with high-complexity Medicare patients. The ROI math is direct: every minute returned to the physician is a minute available for chronic-care management, which is the activity that moves Star Ratings.

Conversational claims and member service. Humana has been one of the more aggressive payers in piloting conversational AI for claims status, prior authorization status, and benefits questions — the three call types that drive the highest member-service volume and the highest member frustration. The pattern here is similar to what we covered in AI for insurance claims processing in 2026: not chatbot deflection, but AI as the first conversational layer that captures intent, then either resolves or routes with full context.

Voice IVR replacement. Traditional IVR ("press 1 for claims") is the worst possible interface for a senior population. Humana's voice-AI investments are aimed at replacing decision-tree IVR with natural-language voice agents that understand "I got a bill I don't think I should have" without forcing the caller through six menu levels. This is the single highest-leverage interface change a senior-first insurer can make.

Predictive readmission and care-gap closure. Humana applies predictive modeling to identify members at high risk of hospital readmission or with open HEDIS care gaps. The model is upstream of the conversation; the conversation is downstream, where a human or AI agent reaches out with a contextual outreach rather than a generic "please get your screening done" email blast.

The unifying theme across all four: AI is being deployed to capture and act on conversational context, not to replace humans with chatbots. That distinction matters. We made the same point in our Lemonade case study on conversational AI in insurance — the winners are the carriers using AI to make conversations possible at scale, not to avoid them.

Social Determinants of Health: The Survey vs. Conversation Gap#

Social determinants of health (SDOH) — housing stability, food security, transportation, social isolation — drive a significant portion of health outcomes for Medicare populations, but they are notoriously hard to capture with checkbox forms.

The problem is one of disclosure, not data collection. A 78-year-old member who is rationing insulin because of cost is unlikely to check a box labeled "I cannot afford my medication." The same member, asked in conversation "how have you been managing your medications this month?" — will frequently disclose the issue in the third or fourth turn of the conversation.

This is exactly the gap we built Perspective AI to close. AI interviews follow up on vague answers, probe for context, and capture the "why now" behind a member's situation. For an MA plan, this is not a research nicety. It is a Star Rating, risk-adjustment, and total-cost-of-care problem in one. A conversational SDOH screen that surfaces food insecurity in turn three is the difference between a $12 produce-prescription intervention and a $14,000 ED visit.

Humana has been one of the louder payer voices on SDOH integration, partly because the company's vertically integrated CenterWell footprint lets it act on what it learns. But the data-capture layer — the part that turns a member statement into a record — is still the bottleneck across the industry. Conversational AI doesn't just collect SDOH data faster. It collects data that forms structurally cannot get.

For payers thinking about how to deploy this, our overview of health insurance AI in 2026 — member engagement, claims, and the compliance reality covers the operating constraints HIPAA-bound conversational deployments need to handle.

The 2026 MA Reimbursement Squeeze: Why AI Becomes a Margin Lever#

Conversational AI inside an MA plan is no longer an innovation experiment. It is a margin defense.

CMS's 2024 and 2025 Advance Notices delivered effective rate cuts and tightened risk-adjustment models (the V28 transition continues phasing in through 2026), with most MA carriers reporting pressure on medical loss ratios and earnings per member. Layered on that: rising medical utilization as the post-pandemic deferred-care surge continued through 2024–2025, and continued scrutiny of supplemental benefit spending.

For Humana, with ~90% MA exposure, this is not a passing cycle. It is a structural reset of unit economics. Every operating dollar has to do more. That has three direct implications for conversational AI deployment:

1. Call center cost per resolved contact has to drop. Voice-AI deflection that resolves benefits questions and claims status calls without human agent time is now a board-level metric, not a CX nicety.
2. Risk adjustment accuracy has to improve. Conversational HRA (Health Risk Assessment) and annual wellness visit pre-work captures more accurate condition documentation than forms. Better documentation = more accurate (and more defensible) risk scores.
3. Member retention has to go up. Voluntary MA member churn is highest among members who felt unheard during a problem moment — the lost claim, the surprise bill, the prior auth denial. Conversational AI that captures and routes those moments is a retention lever.

This is the same logic Aetna, UnitedHealth, and other top-five payers are working through right now — see our companion analyses on Aetna and CVS Health's AI strategy, UnitedHealth Group's AI strategy, Cigna's conversational care navigation, and Elevance Health / Anthem Blue Cross's member-first conversational strategy. The difference at Humana is concentration: there's no commercial book to subsidize a slow transition.

The Lemonade Lesson: What a Senior-First Carrier Can Copy#

Lemonade is a P&C insurer, not a health plan, and its customer base skews young and digital-native. So why is it relevant to Humana?

Because Lemonade is the cleanest worked example of what happens when an insurer makes the conversation — not the form — the system of record for the customer relationship. Our Lemonade case study on conversational AI in insurance lays out the model: AI agents handle intake, claims FNOL, and renewals; humans handle the genuinely complex cases; the customer never types into a 14-field form unless they want to.

For Humana, the translation is not "build a chatbot named Maya." It is "stop assuming the form is the default interface." Specifically:

• Replace HRA forms with conversational HRAs. Annual Health Risk Assessments today are mostly checkbox surveys with 5–15% completion rates among older MA members. A voice-first conversational HRA, embedded in the post-enrollment outreach call, can do double or triple that and capture far richer condition data.
• Replace prior-auth status portals with voice agents. Members don't want to log in to a portal. They want to know if the surgery is approved. A voice agent that answers that one question is more valuable than a redesigned web portal.
• Replace SDOH paper screeners with conversational screening. As covered above — the data only exists in conversation.

The pattern in Branch Insurance's AI-native member experience and Pie Insurance's AI-first workers comp underwriting is the same — the carriers winning are the ones treating conversation as the primary capture mechanism, not the escalation path.

How Perspective AI Fits: Voice-First Conversational Research for MA Populations#

Perspective AI is built for exactly the use case Humana faces every day: capturing the messy, contextual, "why now" answers that forms structurally cannot. Our Interviewer agent runs voice and text conversations at scale, follows up on vague answers, and produces structured output that flows into existing systems. Our Intelligent Intake product is the form-replacement layer for member-facing intake flows.

For a Medicare Advantage plan, three Perspective AI workflows map directly to the business problems above:

• Conversational HRAs and annual wellness visit prep — voice-first, follow-up-capable interviews that capture chronic condition status, SDOH context, and care preferences in member language, not dropdown language.
• Quarterly voice-of-member research — continuous discovery with active MA members to understand churn drivers, plan-design pain points, and supplemental benefit usage at far higher fidelity than satisfaction surveys.
• Disenrolling-member exit interviews — every involuntary or voluntary disenrollment is a chance to capture why. AI interviews scale this from a sampled survey to every exiting member.

Built for CX teams operating under HIPAA-bound constraints, Perspective AI is the modern alternative to Qualtrics-style enterprise CXM platforms that still fundamentally route members into surveys. The case for the switch is even sharper in our broader analysis of AI tools for customer experience in insurance support and why product teams are sunsetting NPS in 2026.

Frequently Asked Questions#

How is Humana using AI today?#

Humana is deploying AI across four primary workstreams: ambient clinical documentation inside CenterWell primary care clinics, conversational claims and benefits inquiries for member service, voice-AI replacement of traditional IVR menus, and predictive modeling for readmission risk and HEDIS care gaps. The unifying theme is conversational AI insurance interfaces that capture context — not chatbots designed to deflect members away from human agents.

Why is Medicare Advantage particularly suited to voice AI?#

Medicare Advantage members are older, more voice-comfortable, and more likely to struggle with traditional web forms than any other US insurance cohort. The median MA member is around 75 years old, and older adults grew up doing business over the phone, which makes them measurably more comfortable with natural-language voice interfaces than with chat or web forms. Voice AI also handles the accessibility and language-complexity barriers — vision, motor, bilingual — that dropdown-based forms structurally fail at.

What is the 2026 Medicare Advantage reimbursement squeeze?#

The 2024 and 2025 CMS Advance Notices delivered effective rate pressure and accelerated the V28 risk-adjustment model phase-in, which together compress per-member economics for MA plans through 2026. For an MA-concentrated payer like Humana, that means every administrative dollar must do more work, which makes voice-AI call deflection, conversational HRAs, and AI-driven risk adjustment accuracy direct margin levers rather than innovation experiments.

What is CenterWell and how does it relate to Humana's AI strategy?#

CenterWell is Humana's payer-owned primary care, home health, and pharmacy business, with more than 300 senior-focused primary care clinics in the US as of 2026. It was built up partly from the home-health assets Humana retained after selling Kindred at Home's hospice business in 2021. CenterWell is where most of Humana's clinical AI investment — ambient scribes, predictive readmission models, care-gap closure workflows — gets operationalized, because the company controls both the payer and the delivery layer.

How does conversational AI improve social determinants of health screening?#

Conversational AI captures social determinants of health data that checkbox forms cannot reach because members rarely disclose food insecurity, housing instability, or medication rationing on a survey. An AI interview follows up on vague answers and probes for context, surfacing the "I've been splitting my pills" disclosure that a form would never catch. For an MA plan, that data is the difference between a $12 produce-prescription intervention and a $14,000 emergency department visit.

How does Humana compare to UnitedHealth and Cigna on AI strategy?#

Humana's AI strategy is more concentrated on Medicare Advantage and senior-care delivery than UnitedHealth's or Cigna's, because roughly 90% of its revenue depends on government programs. UnitedHealth's Optum business gives it broader exposure across care delivery, pharmacy, and analytics, while Cigna's Evernorth focuses heavily on pharmacy benefit management and behavioral health. All three are investing in conversational AI for member service and claims, but Humana's incentives are unusually sharply pointed at the senior-population voice-AI use case.

Conclusion#

Humana is the cleanest test case in US health insurance for what conversational AI insurance interfaces can actually do. The concentration on Medicare Advantage, the CenterWell delivery footprint, and the 2024–2026 reimbursement squeeze all push the same direction: forms are no longer adequate as the system of record for the member relationship. Voice-first conversational AI — for HRAs, claims, SDOH screening, and prior auth status — is becoming the default member interface, not a pilot.

For payer teams thinking about how to operationalize this, the starting point is not "deploy a chatbot." It is "find the highest-friction member moment in your operation and replace the form with a conversation." That's the work Perspective AI was built for. Run a customer interview with a sample of MA members on your highest-volume call type, and you'll see — within a week — what your current form-based system has been missing for years.