Photo: MTStock Studio/Getty Images
A few months ago, I was advising the CEO of a healthcare software company on product strategy. Part of the work involved interviewing their customers, many of them senior executives at provider organizations. More than once, I heard some version of this, "Yeah, but soon AI will be so good it'll just solve this for us. We won't even need the product anymore."
I was shocked. Not because AI couldn't eventually do what that product does – it might – but building a product that solves a complex workflow well is genuinely hard. Nobody was questioning the assumption that AI would just arrive and make it disappear, or asking who would figure out which problems to point it at, or how to wire it into the workflows and processes specific to their business.
That's the gap I want to talk about. Not the technology gap. The application gap. I'll show where healthcare AI actually stands today, what other industries have already figured out, and one move to make this quarter that could produce a working prototype for a problem an organization has always known was worth solving but never had a cost-effective way to fix.
The numbers confirm the gap
The executive mentioned above is not alone. Eighty-three percent of healthcare C-suites are piloting generative AI, but fewer than 10% are investing in the infrastructure to actually deploy it, and "piloting" usually means an AI scribe or RCM and risk adjustment tooling. Only 35% of proofs of concept ever reach production. Meanwhile, digital health startups raised $14.2 billion in 2025 (up 35% year-over-year).
Implementation has not caught up to the hype, and even where it has, it is not focused on what I think is the simplest, most near-term AI opportunity that almost nobody is talking about: hiring one person, arming them with AI-enabled tools and pointing them at the unglamorous operational problems unique to your business.
The opportunity hiding in the mess
Healthcare has been slow to adopt technology for decades, and that slowness has left behind an enormous backlog of operational problems. For example, only 35% of prior authorizations are fully electronic. The rest rely on fax, phone, or portal-based manual workflows. Provider credentialing often takes up to 100 days and is still largely manual. 25% of total U.S. healthcare spending goes to administrative services.
And every organization has its own version of the problem that no one has fixed: The referral coordination process that breaks across two EHR instances; double data entry between systems that don't talk to each other; denial management patterns unique to your specific payer contracts; staffing logic that lives in one person's head and doesn't scale.
It is not that solutions cannot exist; sometimes they do and organizations don't implement them. Sometimes the problem is too specific to any one business for a vendor to build a product around. Often, the return on investment (ROI) just never justified the cost of custom development, so the problem persisted year after year.
That calculus is changing, and what is happening outside healthcare shows just how fast.
What other industries have already figured out
I bring up other industries not because healthcare is the same, but because they are a useful preview of what is possible for innovative healthcare organizations in the next 12 to 24 months. For the rest, maybe 5 to 10 years.
Eighty percent of software developers now use AI coding tools. The head of product for one of those tools, Claude Code at Anthropic, has not edited a single line of code by hand since November 2025. He still ships dozens of code changes daily. He directs; the AI writes. Shopify CEO Tobi Lütke told his entire company, "Before asking for more Headcount and resources, teams must demonstrate why they cannot get what they want done using AI."
AI-native startups are reaching milestones that took traditional software companies five to seven years and 200-person teams to hit.
The cost of building custom solutions has collapsed. Problems that used to require a $500K budget and a six-month timeline can now be prototyped by one person in a week and deployed within a couple of months. Think about that in the context of all those healthcare problems just described: the referral workflow, the credentialing bottleneck, the double data entry. They just became worth fixing.
So what would I actually tell that exec?
If he'd asked me for strategic advice, I'd tell him two things.
First: AI can absolutely solve many of the problems your business faces, but it will not happen on its own. You need someone who understands the company's operations deeply enough to identify the highest-impact problems and build solutions for them–someone who knows healthcare ops and is fluent enough with AI tools like Claude Code and Codex to actually build what's needed. That person probably does not exist in your organization today. Start looking, because they will be increasingly in demand.
Second: once you have them, do not point them at problems you already have a vendor for; point them at the messy, unglamorous side of the company that has been a drag on your business for years–the problems where solutions exist in theory but the ROI never justified the cost. That math is different now. Give them a HIPAA-compliant AI API token, cloud infrastructure with a BAA and a list of five problems. They will "ship," which is what the tech world calls it when going from an idea to a working product – and they will do it fast.
A working prototype in weeks. A deployed solution in months. The provider and payer organizations that operate this way will outperform those that do not on every metric that matters: cost, patient experience, provider satisfaction and quality.
That CEO's customers were half right; AI will eventually reshape what their product does but it will not do it by itself. Somewhere between the hype and real impact is a person with the right tools, pointed at the right problem. The healthcare organizations that find that person first will be the ones that pull ahead in the next 12 to 24 months.
About the author Richard Mancuso is the founder of Insight Healthcare Advisors, a healthcare management consulting firm that advises provider organizations, community-based organizations and PE-backed healthcare companies on Medicaid strategy, commercial growth and AI-enabled operations. He is a guest lecturer at the UC Berkeley School of Public Health. |
Image
|
