UnitedHealth Group is deploying artificial intelligence to call doctors' offices, schedule patient appointments, summarize medical charts on nurses' drives, and analyze millions of customer calls - part of a $3 billion AI investment planned for 2026 and 2027. The insurer, the largest in the U.S., is betting the technology will turn around a steep profit slump last year while cutting the $80 billion that insurers and medical providers spend annually on administrative transactions.
Executives said the push is already delivering a 2-to-1 return. AI automates manual processes such as prior authorization checks and customer service analysis, and UnitedHealth expects to reduce operating costs by nearly $1 billion this year largely driven by AI. "The cost savings potential is clear, particularly for manual, data-intensive processes such as prior authorization," Morgan Stanley analysts led by Erin Wright said in a June research note, adding that UnitedHealth will also sell AI products and services to other healthcare companies.
Automating healthcare's administrative load
An early system called Optum Real lets medical providers check in real time whether a service is covered. It has processed about a billion transactions since launching last year. Other applications read aloud chart summaries as nurses travel to patients' homes, while a trial has AI agents calling doctors' offices directly to book appointments.
UnitedHealth has over a thousand AI uses, 20,000 AI engineers, and 117 large language models available to staff, said Sandeep Dadlani, CEO of the company's Optum Insight data and technology division. Still, Dadlani said almost 99 per cent of AI applications are administrative rather than clinical. "We are not getting into diagnostic AI," he said.
Wall Street has responded. Shares of UnitedHealth have risen 21 per cent this year after falling by more than a third in 2025. The company's insurance division, now led by Tim Noel following the 2024 killing of former chief Brian Thompson, has been under public and political pressure to ease prior authorization requirements.
Public mistrust and litigation
Trust remains a significant obstacle. A Gallup survey last year found 69 per cent of respondents had little or no trust in businesses to use AI responsibly. A separate KFF poll showed about half of insured Americans encountered barriers to care including delays and denials. UnitedHealth faces class action lawsuits claiming it relied on an algorithm to limit post-acute care admissions. The company disputes the claims and says the algorithm wasn't used to make coverage decisions.
A federal inspector general report found that companies using the algorithm, from UnitedHealth subsidiary naviHealth, had higher denial rates, and those denials were almost always overturned on appeal. UnitedHealth said the algorithm does not dictate how long patients can stay in care facilities and that it is used appropriately.
Noel acknowledged the challenge. "You have to gain trust, earn trust through your actions," he said. "The changes will take some time for that to actually be felt by people."
UnitedHealth's guardrails for AI
An internal review board that includes medical ethicists, clinicians, technologists, and privacy and legal experts must clear new AI uses. Dadlani said the company occasionally pulls back a model if it shows unintended behavior. "We have all our alerts firing if something's beginning to drift," he said. The company is also tracking whether some workers use AI at least once a day as it pushes to embed the technology into operations, though Dadlani indicated no direct correlation to job cuts.
This week UnitedHealth invited reporters to its Minneapolis headquarters - an unusual step for a company that historically limited press engagement - to demonstrate AI work and make the case that the technology benefits patients, not just the bottom line. "We only approve using AI," Dadlani said. "We never deny using AI."
The use of AI for insurance processes like claims analysis and prior authorization is accelerating, while AI for healthcare applications remain firmly administrative under the company's review framework.
Why this matters for healthcare professionals
UnitedHealth's scale means its AI choices will ripple across the sector. For nurses, care coordinators, and administrative staff, AI is already handling chart reading and scheduling - changing daily workflows without, so far, replacing jobs. For clinicians, the emphasis on administrative rather than diagnostic AI matters: tools like Optum Real could reduce the phone-and-fax friction of prior authorization, but the algorithm's role in coverage denials continues to face legal scrutiny. Watching how UnitedHealth's internal review board operates and where models are pulled back offers a real-world test of safety guardrails at massive scale. Professionals who understand these patterns will be better positioned to work alongside AI systems and advocate for their responsible implementation.
Your membership also unlocks:
