Summary: UnitedHealthcare rolled out Avery on March 26, 2026 — a generative AI companion members reach through the UnitedHealthcare app and myuhc.com. Notably, the announcement never names the model providers underneath, which says something on its own: enterprises now treat model choice as an implementation detail rather than a marketing hook. The opening wave spans roughly 6.5 million members on employer-sponsored plans along with 160,000 Medicare Advantage members, and the company intends to extend coverage to 20.5 million commercial, Medicare, and Medicaid members before 2026 closes.
UnitedHealthcare rolled out Avery on March 26, 2026 — a generative AI companion members reach through the UnitedHealthcare app and myuhc.com. The opening wave spans roughly 6.5 million members on employer-sponsored plans along with 160,000 Medicare Advantage members, and the company intends to extend coverage to 20.5 million commercial, Medicare, and Medicaid members before 2026 closes.
Read as a product story, Avery amounts to a stress test of unusual scale: generative AI installed at the front entrance of American health insurance. UnitedHealth Group, the parent company, plans to channel $1.6 billion into AI this year, spread across more than 1,000 AI initiatives — and Avery, though just one of them, is the one standing face to face with millions of paying members, where there is the least room for error. Notably, the announcement never names the model providers underneath, which says something on its own: enterprises now treat model choice as an implementation detail rather than a marketing hook.
Avery’s Job Description: The Insurer’s AI Reception Desk
The workload Avery picks up is the question set members actually bring every day: what coverage and personal benefits include, appointment booking (it will phone in-network primary care providers on a member’s behalf), cost estimates and plan balances, rewards and wellness programs, ID cards and over-the-counter benefits, and where a claim stands in approval together with its explanation of benefits. When the task is finding care, Avery leans on UnitedHealthcare’s long-standing Smart Choice provider search. Every interaction teaches it something, and its answers shift to fit one person’s benefits and demographics. By the announcement’s account, the design goal is a simpler, more coordinated care experience — and the personalization is not members-only: advocates, brokers, and employers are all meant to get tailored experiences running on the same rails.
The Figures That Count: 90% Never Reach a Human
- In service today: roughly 6.5 million employer-sponsored plan members, with 160,000 Medicare Advantage members on top
- Where this heads by end of 2026: 20.5 million members across commercial, Medicare, and Medicaid
- The company’s own count: 90% of the time, members working with Avery ended up not needing a live advocate
- $1.6 billion set aside for AI across UnitedHealth Group this year; of its 1,000+ AI solutions, Avery is the newest
- Underneath the product: insights accumulated from billions of digital transactions
Handing Off to People: Full Synopsis, Nothing Repeated
A conversation Avery cannot finish gets auto-routed to a live advocate with a full synopsis attached — the member neither restates the problem nor uploads documents a second time. On the advocate’s side, real-time insight into member history trims the note-taking and the hunt for information, so their work resumes exactly where Avery stopped rather than back at zero. Dan Kueter, CEO of UnitedHealthcare’s commercial business, pitched the launch the way a consumer brand would: people want health care that is easier to use and tailored to their needs, and Avery lets members “focus on what matters most: getting and staying well.”
Oversight, and Where the Risk Sits
Errors carry a steep price in health insurance. On risk, the official position is that Avery runs inside rigorous oversight and governance frameworks covering safety, fairness, and privacy, with human expertise steering the output. The hard questions live in the details: when a cost estimate turns out wrong, who answers for it; whether care guidance quietly stands in for clinical judgment; and of the conversations the AI screens out — that 90% — how many belonged with a person from the very start. Every step faster in the rollout raises the pressure for verifiable answers.
What Product Teams Can Take Away
- Enterprise AI is moving from “assistant” toward “institution”: with Avery, the experiences of members, advocates, brokers, and employers get rebuilt at once — not a chat window bolted onto an app
- “90% resolved without a human” makes a north-star metric worth tracking, but only read next to satisfaction and complaint rates — alone, it turns into cover for trimming service costs
- “AI front door plus human back office plus a complete handoff synopsis” describes a support architecture other teams can copy; a transfer that never forces the user to start over is exactly what sets the trust floor for AI-fronted service
- Keep an eye on the Medicare and Medicaid expansions through the rest of 2026: older and lower-income populations will show whether plain-language AI guidance genuinely holds up