Every night, inside more than 50 U.S. hospitals, an AI system called Healthleap quietly reads through every adult inpatient’s chart — the clinician notes, the lab values, the vitals, the medication orders — looking for a handful of words that are easy for a busy care team to miss: unintended weight loss, trouble swallowing, reduced appetite. By morning, it has written a risk score directly into that patient’s electronic record, flagging who might need a closer look. The San Francisco-based startup announced on October 8, 2026 that it has raised $38 million, split between an $8 million seed round co-led by Sequoia Capital and First Round Capital and a $30 million Series A led by Hummingbird Ventures, to expand a platform that started as a tool for dietitians and has grown into a broader bet on AI-driven diagnosis of conditions hospitals routinely overlook.
A sibling-founded company built around one stubborn statistic
Healthleap was founded in South Africa in 2022 by siblings Jemima and Josiah Meyer. Jemima, a clinical nutrition specialist, originally built a tool to help dietitians track malnutrition risk; Josiah, now the company’s CEO, helped turn it into a platform that plugs directly into a hospital’s electronic health record system and uses language models to extract clinically meaningful signals from free-text notes that would otherwise sit unread in a chart. The company’s pitch rests on a genuinely striking gap in hospital care: research cited by the company and reported by TechCrunch suggests that somewhere between 20 and 50 percent of hospital inpatients are malnourished to a degree that’s clinically significant, yet only a small fraction are ever formally diagnosed. Malnutrition of this kind is linked to longer hospital stays, impaired wound healing, higher infection rates, and elevated mortality — but because it rarely announces itself as dramatically as an abnormal lab value, it’s the kind of condition that can slip past even attentive clinicians managing dozens of patients.
How the system actually works
Healthleap’s core bet is that most of the predictive signal for these overlooked conditions already exists inside the hospital’s own records — it’s just scattered across narrative text that no one has time to systematically mine. The platform ingests notes, labs, vitals, orders and problem lists, uses AI to pull out relevant symptom mentions, and feeds those alongside structured data into risk models. Crucially, the company is explicit that it does not diagnose patients itself; it surfaces cases for a clinician to review, keeping a human in the loop for the actual medical judgment. The malnutrition module is Healthleap’s most mature product, but the company says programs targeting delirium, aspiration pneumonia, pressure ulcers, and congestive-heart-failure readmission risk are now undergoing clinical validation, with an eventual goal of covering more than 40 conditions.
Early results, with the caveat that they’re company-reported
Healthleap’s customer roster already includes Penn Medicine, Houston Methodist, Cedars-Sinai, Intermountain Health, and Emory Healthcare — a jump from just three hospital partners a year ago to more than 50 today, alongside revenue the company says has grown more than tenfold over the same period. At the Hospital of the University of Pennsylvania, the company reports its malnutrition program generated $23.8 million in annualized financial impact: $6.3 million from additional reimbursement tied to properly documented diagnoses, and $17.5 million attributed to shorter hospital stays. The company says every customer has seen at least a 5x return on investment, with some exceeding 20x. Sequoia partner Alfred Lin, explaining the firm’s bet on the company, said Healthleap "is focused on finding patients who are being missed and getting them the care they need."
It’s worth being clear-eyed about those numbers: they are Healthleap’s own figures, self-reported to reporters rather than independently audited, and TechCrunch’s coverage noted it could not independently verify how the Penn impact figure was calculated. ROI claims from a vendor mid-funding-round are, almost by definition, the most flattering version of the story available. That doesn’t make them false, but it’s reason to treat them as a starting point for due diligence rather than settled fact.
The bigger pattern: AI as a second set of eyes on data nobody reads twice
Healthleap’s rise fits into a broader, well-funded category of health AI that isn’t about replacing physicians with chatbots or reading radiology scans, but about mining the messy volume of text hospitals already generate and rarely revisit. Clinical notes are written primarily for billing and communication between providers — never designed to be systematically re-analyzed for patterns a tired clinician might miss on day three of a five-day admission. Investors backing this round are betting that language models are now reliable enough to do that re-analysis cheaply, at scale, across an entire hospital’s patient population every night, something impossible to do manually.
What skeptics will want to watch
The obvious risk with any AI system that flags more cases for review is alert fatigue: too many low-value flags could push stretched nursing and physician staff to reflexively dismiss them, undermining the premise of catching what humans miss. There’s also a harder question beneath the expansion plans — malnutrition detection is a relatively contained, well-studied clinical problem, but the four additional conditions now in validation each carry their own diagnostic subtlety, and it remains unclear whether one platform architecture generalizes cleanly across all of them.
What’s next
With the fresh capital, Healthleap plans to expand engineering, sales, and customer-success teams and push toward its 40-plus-condition roadmap, while eyeing outpatient and home-care settings beyond the hospital walls where it currently operates. Pricing is structured around three-year contracts tied to a hospital’s licensed bed count plus an outcomes-based component, aligning revenue with the financial case made to hospital CFOs. The real test won’t be another funding round — it will be whether independent, published outcomes data holds up as well as Healthleap’s own figures suggest, and whether flagging more patients genuinely changes what happens to them clinically, not just what gets billed.
Photo: fernandozhiminaicela / PIXABAY via Pixabay
