Before Eli Ben-Joseph built software for doctors, he tried to decide whether he should become one. The question followed him out of MIT, where he studied biology and bioengineering, and through a first job in consulting. He liked technology. He liked making things. He did not much like the sensation that consulting was someone else's answer to his own life. Medical school seemed more useful, more direct, and connected to a family example: his grandfather had been a physician.
Then came the inconvenient detail. In hospitals, Ben-Joseph saw doctors spending a startling amount of their day negotiating with computers. During one period of shadowing, he recalled a ratio of 15 minutes at a screen for every minute with a patient. The electronic record, supposedly the great cabinet of clinical memory, had become a cupboard so crowded that opening the door risked an avalanche.
He changed the question. Instead of asking whether he should practice medicine, he began asking what sort of machine might make medical practice more human. It was a useful reversal. It would also consume the next decade.
First, resist the clever answer
Ben-Joseph went to Stanford for a master's that joined management with computer science. The timing was 2014 to 2016, when deep learning had returned to fashion and every ambitious graduate could glimpse a dozen possible futures in it. He reconnected with longtime friends Nate Wilson and Thomas Moulia. Together they began imagining what a doctor with genuinely useful software might look like.
They did not begin by naming a feature. They spent roughly eight months talking to hundreds of doctors, nurses, and healthcare executives. Admission to the Techstars Cedars-Sinai accelerator gave them a close view of hospital life. They watched. They asked questions. They learned that clinical work does not suffer from a shortage of data. A patient record may contain tens of thousands of data points. What is scarce is the time to find the few that matter now.
The company, founded in 2016 as HealthTensor, spent much of the next two years building and testing. Its premise was simple enough to fit on an index card: review the record, find possible diagnoses a busy clinician might not see, and place those suggestions inside the workflow. The clinician would make the decision. The software would do the tireless reading.
“Adoption moves at the pace of trust.”Eli Ben-Joseph, at the 2025 Regard Summit
That sentence has the clipped neatness of a founder's maxim, but it also describes the drag coefficient of healthcare. A consumer app can be forgiven for a foolish notification. Software that enters a hospital has to earn a different kind of permission. Ben-Joseph learned this with Regard's first major customer. The opening was neither a warm introduction nor a theatrical demo day. It was a cold email to a medical director at Torrance Memorial Medical Center.
The director replied. What followed was a year of collaboration with physicians, testing the product and refining it from their feedback before a contract was signed. A cold email had opened the door; patient work, in both senses of the adjective, kept it open.
The founder who thinks he moved too slowly
Ben-Joseph has said the team made an early mistake by hiring and fundraising too slowly. Self-doubt encouraged caution. They wondered whether hospitals would truly want what they were building. He has also said that the company got its product discovery right because it refused to hurry product-market fit. Regard has never had to pivot.
These statements are not contradictory. A founder can be too hesitant about assembling a company and exactly patient enough about deciding what that company should make. The trick is telling the two forms of caution apart while living through them, when both feel like prudence and both send invoices.
In 2022, HealthTensor became Regard, a name with unusually good manners for a technology company. That year it raised a $15.3 million Series A. In July 2024, Regard announced a $61 million Series B led by Oak HC/FT, with Cedars-Sinai Health Ventures and existing investors participating. The money was designated for product work, large-language-model research, and expansion beyond inpatient facilities.
The numbers attached to adoption began moving faster too. The first million diagnoses surfaced by Regard took two and a half years. The second million took 11 months. Ben-Joseph received that particular update while attending a Lakers-Celtics game on Christmas Day, a tidy collision of rivalry, holiday, and dashboard. By 2025, he reported that clinicians had accepted more than 10 million diagnoses surfaced by Regard.
An inevitable future, built inconveniently
Ben-Joseph and Wilson use the phrase “inevitable futures” to think about strategy. Renewable energy is one example: the destination can be visible while the route remains full of engineering, policy, and commercial potholes. Ben-Joseph puts computer-assisted clinical reasoning in the same category. In his view, computers will take a larger role in making sense of records and personalizing care. The interesting question is not whether they will be present, but how they will behave once admitted.
His answer is more practical than cinematic. Regard reviews the full longitudinal record, recommends possible diagnoses, and drafts clinical documentation for a doctor to assess. Its newer Proactive Documentation product combines information already in the record with the conversation in the room. An AI agent called Max can answer questions about the chart and help prepare notes. In 2026, Regard demonstrated how its clinical intelligence could work with Microsoft Dragon Copilot, joining ambient capture with the longer memory of the record.
This approach puts Ben-Joseph in a delicate position. He has to argue that AI can materially change a clinician's work without pretending that medicine is merely an untidy information-retrieval problem. He often returns to augmentation. His sharper formulation is that AI will not replace doctors, but doctors who use it may replace those who do not. Beneath the provocation sits an ordinary observation: professions change when their tools change.
The market has rewarded the argument. By 2026, Regard said its software was used across more than 150 hospitals. Fast Company placed it eighth on its 2026 healthcare innovation list, citing deployment in 15 states and company-reported improvements in diagnosis and documentation. Ben-Joseph was selected for Crain's New York Business 40 Under 40. Regard also began extending its platform into cardiology, surgery, outpatient encounters, case management, quality, and clinical documentation improvement.
The ambition behind the modest interface
For a founder interested in large outcomes, Ben-Joseph is drawn to unglamorous bottlenecks. His Stanford machine-learning work included a project that predicted wine varietals from professional reviews and another that explored guiding a viewer's gaze with subtle changes in illumination. Both projects involved extracting a useful signal from perception and language. Regard applies the pattern where the stakes and institutional resistance are much higher.
He is candid that the work could fail. When asked about the hardest, most rewarding goal he could pursue, he chose Regard itself. Stanford, he said, was a decision to seek massive impact, and healthcare offered the broadest field for it. His stated horizon extends beyond American hospitals to anyone who can reach a computer and the internet.
That global aspiration is far ahead of the current footprint, which remains national. Yet it clarifies why he keeps talking about the whole record. The immediate product saves a doctor from hunting through a chart. The larger thesis is about access to expertise: if clinical knowledge and individual history can be compressed into something intelligible at the right moment, geography matters a little less.
“If you can create a system that allows 100% of that data to be reviewed, you can really amplify the clinical abilities of the doctor.”Eli Ben-Joseph
There is a pleasing irony in the career choice. Ben-Joseph stepped away from becoming a doctor after watching medicine become too computerized. Then he founded a computer company. The distinction is in what he wants the machine to do. The bad computer demands attention. The good one returns it.
Regard still has to prove that distinction hospital by hospital, specialty by specialty, and workflow by workflow. Trust, in Ben-Joseph's phrase, sets the pace. The data is already there. The difficult work is making it useful without making the clinician serve the tool. A decade into the attempt, the company's most important design principle remains the observation that started it: somebody should read the whole chart, and it does not always have to be the exhausted person in the room.