Before we get to this week’s column—Please consider joining us at the inaugural Fortune AIQ Summit at the New York Stock Exchange on Oct. 1: Spend the afternoon with senior executives from companies on the Fortune AIQ 75 list and explore how you can scale your AI experimentation and translate investments into measurable business value. Apply here to attend. After Katya Andresen joined Cigna Group as chief data, digital and AI officer in September 2021, she says the emergence of generative artificial intelligence as a disruptive new technology led many to ask themselves: “How do I use AI?” Andresen thinks that’s the wrong question. “We’ve been on a mission to change that question to, ‘how do I lead in an age of AI?’” says Andresen. That framing is less focused on the functionality of any given AI tool and chasing countless use cases, and instead homes in on measurement that shows an AI investment can change health outcomes.Along those lines, Cigna has announced this summer that it projects the AI and predictive analytics tools that the company uses to help patients identify chronic conditions—including cancer, kidney disease, and high-risk pregnancy—can save an estimated $200 million over the next three years by proactively connecting patients with clinicians. Separately this summer, Cigna announced a $100 million investment through 2028 to use AI to reduce the time clinicians need to spend documenting their cases and speed up the prescription process.And yet another AI use case, which Andresen shared with Fortune, involves using AI to better understand common inbound patient questions that Cigna was receiving about biosimilars, which can treat chronic diseases at a fraction of the cost of biologics. Unlike a generic drug, which has the same active ingredients as branded counterparts, biosimilars aren’t an exact match because they’re made from living organisms like bacteria and plant cells. But both are approved by the Food and Drug Administration.One biologic, called Humira, can cost a patient $7,000 per month to treat inflammatory and autoimmune conditions. But Cigna looked at thousands of prior customer conversations with its representatives about biologics and biosimilars, and then used those insights to craft stronger digital messaging to encourage a switch to the cheaper alternative. This targeted campaign led more than 80% to opt for the biosimilar, Andresen says.“That led to a lot more margin,” adds Andresen. “But more importantly, it created a couple hundred million dollars of savings for patients.”Extracting millions in cost savings from these high-priority use cases is critical for Cigna, which generates $275 billion in annual revenue and ranked 14th on the latest Fortune 500 list, as national healthcare spending exceeds $5 trillion annually due to the rise of chronic conditions, an aging population, and the cost of a hospital stay and average prices for new prescription drugs soaring.Patients frequently express that they are fed up with navigating the industry’s complex system, and millions report they’ve turned to ChatGPT and other AI chatbots to ask health- or healthcare-related questions. Nearly six-in-ten report using AI to research health information before a doctor visit, and about 14 million adults say they have skipped a visit with a provider after using AI, according to a survey published by Gallup in April.This movement does raise thorny questions about the guardrails put in place around AI chatbots, even those created by insurance and pharmaceutical companies, because they need to handle sensitive patient information and have the ability to accurately answer complex questions about medical insurance plans and treatments.“The good news is, because we are highly regulated, we have a massive amount of controls in place to begin with,” says Andresen. “We are not starting from zero.” That layer of compliance and governance has existed for well over a decade for the machine learning models that Cigna has leveraged, she adds, and is also closely controlled for any data that’s access by third-party vendors.Andresen has had to recently hunt for answers to these common healthcare questions after a close family member was diagnosed with breast cancer. She says this experience has shown that personalization, not just navigation, is the differentiator that AI can provide.“I think we’re headed to a place where we are going to find that AI in healthcare becomes conversational, ambient, and more proactive,” says Andresen. “We can be more and more precise with treatments, with recommendations, and we can get better and better at understanding what works…and feed that back into our models, so that everything gets better all the time.”Five years ago, when Andresen joined Cigna, it was her first role leading a healthcare company, after previously serving as a senior vice president of financial services at Capital One, serving in executive leadership roles at mission-focused tech firms Cricket Media and Network for Good, and earlier in her career, working as a foreign correspondent for Reuters News and the Associated Press.She says that AI technology is evolving so rapidly that a mix of Cigna’s own proprietary data to build competitively specific tools, as well as key partnerships with big AI players, will be the differentiator. Andresen has launched workplace tools like chatbot Microsoft Copilot and AI coding agent Cursor, while also working closely with large hyperscalers like OpenAI and Anthropic to tap their large language models.There are also times when Cigna will opt to work closely with AI startups that have homed in on a very specific use case, like Sierra, which builds conversational AI agents for customer service. “We’ve worked really closely with them on their product roadmap,” says Andresen.Other generative AI use cases that Cigna has deployed include using LLMs to summarize millions of phone calls placed to call center agents, and then leveraging those insights to create an AI tool that makes it easier for those employees to search for the answers to questions like, “does my policy cover this treatment for plantar fasciitis?”An AI virtual assistant was also built inside Cigna’s mobile app, a conversational tool that can similarly address patient questions, while in the clinical setting, AI-enabled summarization has reduced note-taking by up to 90% for health practitioners who work for Cigna’s telehealth service MDLIVE.“The principles behind all this are hopefully clear, which is, what problem are we trying to solve, and how can AI help?” says Andresen. “That’s always the starting point.”John KellThis story was originally featured on Fortune.com