A San Mateo company that started with text-message reminders is now the operational AI layer for 650 hospitals - and the patient on hold is starting to notice.
Walk into a midsize health system in 2026 and the first thing you notice is the silence at the front desk. Not the dystopian kind. The relieved kind. Phones still ring, but fewer staffers lunge at them. The fax machine still hums - a small concession to American medicine's stubborn analog backbone - but its outputs now route themselves. Somewhere, a patient is rescheduling a cardiology visit at 11:47 p.m. without ever speaking to a human. That patient is talking to Luma.
Luma Health is a 190-person company in San Mateo that has spent a decade gluing together the most reluctant software stack in technology. Hospitals run on EHRs - electronic health records - and the EHR is a famously unfriendly neighbor. Luma's product is what happens between EHRs, calendars, billing systems, and the people trying to get an appointment. The company calls it a Patient Success Platform. The new marketing calls it Operational AI. The clinics call it the reason their inbox is quieter.
In 2014, Adnan Iqbal sat in a waiting room for a routine appointment and watched two phenomena collide. The waiting room was full. The schedule, behind the curtain, was not. Empty 2 p.m. slots stared back at staff who could not, for love or money, get a callback from the patient who needed exactly that 2 p.m. slot. The mismatch was not a bug. It was the system.
American healthcare has roughly 15 to 30 percent of clinic capacity walk out the door each day as no-shows and last-minute cancellations. Meanwhile the average specialist appointment wait time pushed past three weeks. Patients blamed clinics. Clinics blamed patients. Software vendors blamed the EHR. The EHR blamed the fax machine. Everyone was - in fairness - partially correct.
Iqbal teamed up with Dr. Tashfeen Ekram, a radiologist who had spent years watching the same scheduling tragedy from the clinical side, and Aditya Bansod, a product engineer who had spent his career building infrastructure for the rest of us. Their first move was unfashionable: they ignored the EHR and built around it. The first Luma product was a text message - the kind a patient might actually read - that asked one question. Can you make it tomorrow at 2?
Critics, of which there were many, called it lightweight. Hospitals, which famously prefer suites to scalpels, agreed. Then the numbers started arriving. No-show rates fell. Waitlist conversions rose. Phones rang less. Word, as it does in healthcare IT, traveled by golf cart and conference panel.
By 2018, Luma had quietly grown into the patient engagement category - a term that, like most healthcare-IT taxonomy, is roughly two-thirds buzzword. By 2021, it had closed a $130 million Series C led by FTV Capital, with 8VC, Cisco Investments, PeakSpan and U.S. Venture Partners joining the cap table.
A patient's first encounter with a health system is, on average, four phone calls long. The Luma platform tries to collapse that to one - or, ideally, zero. Spark, the company's multi-model generative AI core, coordinates the agents and workflows that touch a patient between the moment they need care and the moment they pay for it.
A voice AI concierge that answers patient calls, books appointments and verifies insurance. Yes, it actually picks up.
AI that classifies and routes inbound faxes. Yes, faxes. American medicine still moves roughly 9 billion of them a year.
Self-service forms that update the EHR directly. Clipboard, retired.
Closes the loop between referring providers, specialists and patients - the part everyone forgets.
The unsexy point: every one of these features only works because Luma has spent ten years writing integrations to more than 80 electronic health record systems. The moat is paperwork. The moat is also, occasionally, regulatory. HIPAA, HITECH, MACRA, MIPS - alphabet soup is a category of competitive advantage when you can read all the letters.
Figures self-reported by Luma Health, 2025-2026. Bars scaled for visual comparison.
The most useful test of any healthcare software is whether the people running the front desk would mutiny if you removed it. By that bar, Luma is doing fine. The University of Arkansas for Medical Sciences reports that Navigator, deployed for less than a month, eliminated 98 hours of manual call-center voicemail review. Sun River Health, a New York FQHC network serving more than 250,000 patients, runs intake on Luma. Bingham Healthcare, TriState Health, Maury Regional - the patient on the other end has no idea they are being routed by software named after the Chinese character for light.
The Inc. 5000 listed Luma as one of America's fastest-growing private companies after roughly 900 percent three-year growth. The company has connected, by its own count, nearly a quarter of all Americans to some form of care.
Luma is careful about its language. It does not say patient engagement, the industry buzzword that has come to mean almost nothing. It says patient success - a deliberate borrowing from the SaaS playbook, where success means a measurable, completed outcome. The patient is the customer. The customer should not have to call four times. The customer should not have to fax. The customer should not lose a referral in a dropdown.
None of which is revolutionary, exactly. What is revolutionary is that an enterprise health system might agree.
Healthcare's AI moment will not arrive as a flashy diagnostic. It will arrive as the call you do not have to make, the form you do not have to fill out, the fax you do not have to scan. The flashy stuff - the LLM that reads your MRI - depends on a less flashy thing actually working: the part where you got the MRI scheduled in the first place. Luma is betting that the boring layer is the layer.
It is a defensible bet. The American healthcare system spends roughly $4.5 trillion a year, and a non-trivial fraction of that is bled out in coordination failures - missed slots, dropped referrals, denied authorizations, no-shows. Even modest dents matter. A platform that connects a quarter of Americans to care has already made a few.
Back at the front desk, the phone rings again. Nobody flinches. Navigator picks up. Somewhere a doctor's 2 p.m. slot quietly fills. The fax machine hums, reads itself, and goes back to sleep. This is not the future of healthcare. This is the present, just unevenly distributed - and Luma Health is the company quietly doing the distributing.