Devoted Health's AI Strategy: How the Tech-First Medicare Advantage Insurer Listens to Members in 2026

TL;DR#

Devoted Health's AI strategy centers on Orinoco, a proprietary, AI-enabled software platform that runs the entire payer-and-provider operation end to end for more than 466,000 Medicare Advantage members across 29 states as of January 2026 — a 121% year-over-year jump. Founded in 2017 by brothers Todd and Ed Park, Devoted pairs that platform with human "Guides" and a virtual-first medical group, an approach that helped it land multiple 5-star contracts and put 95% of members in 4-star-or-better plans for 2026. The hard part for any senior-focused insurer is not the technology — it is hearing from members who do not finish web forms or mail-back satisfaction surveys. Fewer than 60% of Americans 65 and older use the internet, and Medicare CAHPS response rates fall off sharply after age 75, so the standard listening tools systematically undercount the population whose experience drives star ratings and retention. Conversational AI interviews, including voice, let Devoted-style insurers reach seniors in plain language, follow up on confusion, and capture the "why" behind a benefit complaint or a missed appointment. This is the core of Perspective AI's argument: AI-first member research cannot start with a web form.

What is Devoted Health's AI strategy?#

Devoted Health's AI strategy is the use of its proprietary Orinoco software platform — which combines advanced data and AI with clinical and service operations — to deliver proactive, prevention-focused Medicare Advantage care at scale while keeping a human Guide in the loop for every member. Rather than buying and stitching together legacy insurance systems, Devoted built one modern platform to run claims, the electronic medical record, care navigation, and member service in an integrated way, then layered AI on top to reduce administrative waste and sharpen clinical decisions.

That design choice is the whole story. Most Medicare Advantage plans are assembled from decades-old core systems that were never meant to talk to each other. Devoted's founders — Ed Park, the company's CEO and a founding engineer and former COO of athenahealth, and Todd Park, the first Chief Technology Officer of the United States — set out to build what Todd has described as an "American healthcare system as a service." For product and customer-experience leaders, Devoted is a useful case study in what tech-first looks like when the customer is an older adult, and where even a sophisticated tech stack still struggles to hear that customer. It mirrors the build-versus-buy lessons in the complete guide to AI-powered customer experience.

Company snapshot: Devoted Health by the numbers#

Devoted Health is a privately held, technology-driven Medicare Advantage insurer that has scaled faster than almost any peer in the category. The numbers below are the load-bearing facts behind its AI strategy.

The star-rating performance is not a vanity metric. In Medicare Advantage, Star Ratings drive the federal quality bonus payments that fund richer benefits, and a large share of the rating is built from member-experience measures — how members rate their plan, their care, and how easy it was to get help. That makes "can we actually hear from our members?" a financial question, not just a research one. Insurers thinking through this connection should review how to build a voice-of-customer program from scratch and the complete guide to voice-of-customer programs in 2026.

Where Devoted Health uses AI and technology today#

Devoted Health uses AI and proprietary technology across five tightly integrated components, so that data captured in one part of the experience is available everywhere else. The five pieces work as one system rather than separate vendors.

• Orinoco — the proprietary platform that serves as the EMR, the billing-and-claims engine, and the connective tissue across every other component. Devoted describes it as a modern platform able to support the entirety of payer and provider operations end to end.
• Devoted Health plans — the in-house Medicare Advantage product, designed around the data and service model rather than bolted onto it.
• Devoted Health Guides — tech-enabled service representatives, framed as a "guardian angel," who navigate members through appointments, benefits, and billing. AI surfaces context; the Guide carries the relationship.
• Devoted Medical — a prevention-focused, virtual-first medical group that complements a member's existing primary-care network with in-home and telehealth care.
• Contracted provider network — local clinicians in every community Devoted serves, connected back into Orinoco.

On the a16z Raising Health podcast episode "AI-Enabled Continuity of Care," Ed and Todd Park framed AI's role as reducing systemic waste and enhancing clinical decisions — continuity across visits, providers, and time, rather than a flashy member-facing chatbot. This is the same continuity logic that work-management and CRM leaders apply to their own customers, as covered in the HubSpot AI customer research breakdown and the Zendesk AI customer strategy analysis.

The senior listening problem: why forms and surveys undercount the people who matter most#

The core gap in Devoted Health's member experience — and every Medicare Advantage plan's — is that the older population it serves is precisely the group least likely to complete a web form or a mailed satisfaction survey. The tooling built to measure experience is biased against the members whose experience matters most for star ratings.

Three data points define the problem:

1. Internet access drops with age. Fewer than 60% of American adults aged 65 and older use the internet, with markedly lower usage among Black and Latino seniors and among the oldest cohorts, according to research published in the National Library of Medicine. A digital-first intake form silently excludes a large share of the panel.
2. Survey response falls off with age. Analysis of Medicare CAHPS nonresponse found that response rates dropped beyond age 75 and ran 7–17 percentage points lower for Asian, Black, and Hispanic respondents, per a study indexed in PubMed. The members who churn or struggle quietly are the least likely to answer.
3. Even mixed-mode surveys lean on the wrong default. CMS only added an initial web mode to CAHPS in 2024 (web-mail-telephone), and the historical overall response rate sat around 49% — meaning roughly half of sampled members never respond at all.

A satisfaction score built on that base is a survey of the members who were already engaged enough to respond. It tells a plan almost nothing about the 78-year-old who got confused by a prior-authorization letter, gave up, and is now a flight risk at the next open-enrollment window. This is the same structural failure that the forms-versus-conversations research case describes, and that the AI versus surveys comparison quantifies: forms flatten people into dropdowns, front-load effort before any value, and break down exactly at the messy, "it depends" moments where the real insight lives.

How conversational AI interviews meet seniors where they are#

Conversational AI interviews close the senior listening gap by replacing the form with a guided, plain-language conversation — by voice or chat — that an older adult can complete the same way they would talk to a Guide on the phone. Instead of asking a member to translate their frustration into a 1-to-5 scale, the AI interviewer asks an open question, listens, and follows up on whatever the member actually said.

For a tech-first Medicare Advantage insurer, that changes what is measurable:

• Voice removes the keyboard barrier. A voice-based AI interviewer agent lets a member describe a billing problem or a transportation barrier out loud, in their own words — no typing, no logging in, no small fonts. Voice meets the population where its comfort already is: the telephone.
• Follow-up captures the "why," not just the "what." When a member says a benefit "didn't work for me," the AI probes — which benefit, what happened, what did you try? That is the difference between a score and a root cause, and it is the gap the beyond-NPS argument keeps surfacing.
• Scale without losing depth. A plan can run hundreds or thousands of these conversations at once after a plan change, a star-survey period, or a spike in disenrollment, then auto-synthesize the transcripts — the model the customer-research-tools stack for 2026 describes.
• Continuous, not annual. Because conversations are cheap to run, listening becomes a rhythm rather than a once-a-year CAHPS event, feeding the kind of always-on program outlined in the voice-of-customer build guide.

A concierge agent can also replace the static enrollment or onboarding form itself — the same form-replacement move that drove results in the Lemonade conversational-AI insurance case study and the Chime AI onboarding story about replacing forms. For seniors, lowering the cost of responding is the entire game.

The Medicare Advantage context: why this matters in 2026#

In Medicare Advantage, the quality of member listening in 2026 maps almost directly to revenue, because member-experience and access measures carry heavy weight in the Star Ratings that determine quality bonus payments. A plan that cannot hear from its hardest-to-reach members is flying blind on the exact measures that fund its benefits.

Devoted Health's own results show the upside of getting this right: landing multiple 5-star contracts and putting 95% of members in 4-star-or-better plans for 2026 is a quality outcome that compounds into richer benefits, easier growth, and lower churn. But the entire MA category is being pushed in the same direction. The peer set is investing in AI-enabled member experience across the board — compare the UnitedHealth Group AI strategy for member experience, the Cigna conversational care-navigation analysis, and the Centene Medicaid member-experience breakdown. Adjacent healthcare players are modernizing intake the same way, from the Teladoc AI telehealth analysis to One Medical's patient-onboarding modernization and the Walgreens conversational patient-experience strategy.

The differentiator will not be whether a plan has AI. It will be whether the AI is pointed at member understanding — capturing barriers, confusion, and trust in members' own words — or merely at automating the next claim. For CX and product teams inside payers, that is a research question first. Perspective AI is built for CX teams and product teams who need that depth without standing up an enterprise CXM deployment.

Frequently Asked Questions#

What technology platform does Devoted Health use?#

Devoted Health runs on Orinoco, its proprietary, AI-enabled software platform. Orinoco serves as the electronic medical record, the billing-and-claims system, and the integration layer connecting Devoted's health plans, its Guides service team, its virtual-first Devoted Medical group, and its contracted provider network. Devoted describes Orinoco as a modern platform capable of supporting the full scope of payer and provider operations end to end, rather than a patchwork of legacy systems.

How many members does Devoted Health have?#

Devoted Health served more than 466,000 Medicare Advantage members across 29 states as of January 2026, a 121% increase year over year. That makes it one of the fastest-growing plans in the Medicare Advantage category. The company was founded in 2017 by brothers Todd and Ed Park and has expanded its footprint steadily from an early base of roughly a dozen states.

Why do surveys and forms fail to reach Medicare Advantage seniors?#

Surveys and forms systematically undercount older adults because the population skews toward people who are less likely to use digital tools or complete mail-back questionnaires. Fewer than 60% of Americans aged 65 and older use the internet, and Medicare CAHPS analysis shows response rates dropping beyond age 75 and running lower among several racial and ethnic groups. The result is satisfaction data drawn mostly from already-engaged members, missing the confused or frustrated ones who drive churn.

How can conversational AI improve Medicare Advantage member experience?#

Conversational AI improves member experience by replacing static forms with guided, plain-language interviews — by voice or chat — that seniors can complete the way they would talk to a service rep on the phone. The AI follows up on vague answers to capture the "why" behind a complaint, runs at scale across thousands of members, and supports continuous listening rather than an annual survey. For plans, that surfaces the access and trust barriers that weigh heavily in Star Ratings.

Does Devoted Health use AI in member care?#

Yes. Devoted Health uses AI as part of its Orinoco platform to combine advanced data with clinical and service operations, reduce administrative waste, and support clinical decisions, with human Guides carrying the member relationship. On the a16z Raising Health podcast, founders Ed and Todd Park framed AI's role as enabling continuity of care across visits and providers rather than as a member-facing chatbot.

Conclusion: tech-first means listening-first#

Devoted Health's AI strategy proves that a Medicare Advantage insurer can be built tech-first — one proprietary platform, AI woven through care and operations, multiple 5-star contracts, and 95% of members in 4-star-or-better plans for 2026. But the same population that makes Devoted's model valuable is the population that abandons web forms and skips mail-back surveys, and no amount of backend AI fixes a listening layer that older adults cannot or will not use. Understanding seniors' barriers, confusion, and trust is what protects star ratings and retention, and that understanding has to start with a conversation, not a form.

That is the heart of Perspective AI's position: AI-first member research cannot start with a web form. Conversational AI interviews — including voice — let payers reach the members surveys miss, follow up on the messy moments, and capture the reasoning behind a score. If you lead CX or product inside a Medicare Advantage plan or any senior-serving organization, you can start a new research study or explore the AI interviewer agent to hear from the members your current tools are leaving out.