AI CLOCK
Sequoia Capital — the firm that backed Google, Apple and WhatsApp — just wrote a $38 million cheque to a startup that started in South Africa with a dietitian's side project. The startup is called Healthleap, and its job is to read hospital charts overnight and tell doctors which patients might be getting overlooked.
That's the headline. But if you run a business — any business, not a hospital — the interesting part isn't the money. It's the playbook. Healthleap's founders did four things that most businesses get backwards when they try to use AI, and each one is something you can copy this week.
What actually happened
Healthleap was founded in South Africa in 2022 by siblings Jemima and Josiah Meyer. Jemima was working as a clinical dietitian when she built an early tool to help dietitians find patients who needed attention sooner. She teamed up with her brother, and what began as a nutrition tool became something much bigger: an AI platform that screens hospital patients for conditions that often go undiagnosed.
The $38 million comes in two parts: an $8 million seed round co-led by Sequoia Capital and First Round Capital, and a $30 million Series A led by Hummingbird Ventures. The company isn't disclosing its valuation.
Here's how the product works. Each night, Healthleap plugs into a hospital's electronic health record system and reads every adult inpatient's full chart — doctors' notes, lab results, vital signs, medications, diet orders, diagnoses. It uses language models to pull meaning from the written notes (things like "poor appetite" or "trouble swallowing"), combines that with the structured data, and each morning writes a risk score into the care team's existing workflow. No pop-ups, no new dashboard to check. Dietitians open the worklist they already use, sort by risk, and start with the highest-risk patients.
And to be clear: the software doesn't diagnose anyone. It flags patients for a closer look, and clinicians make the call. That distinction matters, and we'll come back to it.
The traction is what pulled Sequoia in. Healthleap has grown from 3 hospital partners to more than 50 in the past year. Its customers now include Penn Medicine, Houston Methodist, Cedars-Sinai, Emory Healthcare and UMass Memorial. Revenue grew more than tenfold over the same period, according to CEO Josiah Meyer — though he didn't disclose specifics. It started with malnutrition and delirium, and it has built programs for aspiration pneumonia, pressure ulcers and congestive heart failure readmission risk that are still undergoing clinical validation. The long-term goal: more than 40 conditions, plus outpatient and home care.
Why this matters far beyond hospitals
Forget the funding for a minute. The reason this story deserves your attention is the way these two founders went about it. Four moves, all of them copyable.
They started embarrassingly specific. Not "AI for healthcare." Malnutrition. One condition, in one workflow, for one group of clinicians. Healthleap estimates undiagnosed malnutrition costs the US healthcare system as much as $58 billion a year — the company says up to half of hospitalized patients may be at risk while fewer than 9% receive a formal diagnosis. That's a painfully specific, expensive, measurable problem. Once they owned that one problem, expanding to delirium, pneumonia and 40 more conditions became a natural next step. Most businesses do the reverse: they buy an "AI solution" for everything and fix nothing. Pick one wound, dress it properly, then move on.
They refused to add a tool. This is the move I'd steal first. Healthleap's risk score appears as a column in a worklist the care team already opens every morning. No interruptive alerts, no pop-ups, no new app. MedCity News reported that the AI combines all its signals into a single priority score from 0 to 10 per patient, plus a second column showing whether the risk changed since the last assessment. That's it. The people doing the work changed nothing about their routine — they just started with better information. If you're bringing any tool into your business, ask the same question: does this remove a step, or add one? If it adds one, your team will quietly stop using it.
They sell outcomes, not software. Healthleap signs three-year contracts priced by a hospital's licensed bed count — and uses outcome-based pricing on top. The company says it contractually guarantees multiples of the contract price based on hard ROI that the hospital's own finance team validates. At the Hospital of the University of Pennsylvania, the company says its malnutrition program delivered $23.8 million in annualized financial impact: $6.3 million from additional reimbursement and $17.5 million from shorter hospital stays. Sequoia partner Alfred Lin said one hospital saw 39% more malnutrition cases identified and treated, with $11 million in incremental revenue in a single health system. You don't have to take those numbers at face value — they're the company's claims. But the strategy is the point: when you sell a result instead of a feature, the conversation stops being about price and starts being about proof. If you sell anything, start measuring the result your customers get, not just the thing you deliver.
They built it in Africa and sold it to the world. This one matters to us. A South African-founded team, now based in San Francisco, is screening patients in more than 50 US hospitals. The money flowed from Sand Hill Road to African-built AI, not the other way around. It punctures the tired idea that serious AI gets built elsewhere and merely deployed here. African founders can build at the top of the stack — and when they do, the market doesn't ask where the code was written.
The wave is already reaching Africa
Meanwhile, the same shift is landing on African soil, in forms closer to our reality.
Rwanda is piloting AI tools in dozens of public clinics to ease health worker shortages — AI-supported systems for patient triage, symptom assessment and medical documentation. The country has roughly one healthcare worker for every 1,000 patients, far below the WHO's recommended four per 1,000, and officials stress the tools are decision support: final diagnoses stay with trained professionals. In Kenya, private clinics are running diagnostic imaging AI alongside doctors — the software flags fractures and pneumonia faster than manual review, and a radiologist still makes the final call. Nigeria's AEyeCare built an AI retinal screening tool that checks children for severe malaria in about 20 minutes, designed for bedsides and clinics rather than big hospitals.
The pattern is identical everywhere: the AI doesn't replace the clinician. It reads faster, flags earlier, and hands a ranked list to a human who decides. That's the shape of useful AI in 2026 — not a robot doctor, a tireless assistant.
What to do this week
You don't need $38 million, a hospital, or a sibling co-founder. Here's the Healthleap playbook, translated for any business:
1. Find your "undiagnosed condition." Every business has a problem that quietly costs it money while everyone stays too busy to notice — abandoned carts nobody follows up, repeat complaints nobody logs, no-shows nobody tracks, stock that expires in the back. Name one this week. Put a number on it, even a rough one. That's your malnutrition.
2. Start embarrassingly narrow. One workflow, one tool, one metric. Healthleap did malnutrition before it touched 40 other conditions. If you try to automate everything in October, you'll automate nothing by December. One win you can measure beats five experiments you can't.
3. Fit it into the workflow people already use. Don't buy a new dashboard. Put the answer where your team already looks — the WhatsApp group, the morning spreadsheet, the order book. Healthleap's genius was a column in an existing worklist. The best AI feature is the one nobody has to open a new tab for.
4. Measure in money, not vibes. Before you start, write down the one number you'll check in 90 days: hours saved, recovery rate, repeat sales. Healthleap sells hospitals on reimbursement and shorter stays, not on "AI-powered insights." If you can't name the number, you don't have a project — you have a hobby.
5. Look African first. The talent that built Healthleap started here. Before you buy a foreign tool or hire a foreign consultant, ask who's building the local version. The continent's AI story isn't "catching up" — in pockets like this one, it's leading. Your vendors, partners and hires should reflect that.
Two dietitians' siblings spotted a condition nobody was looking at, built AI that fits inside a morning routine, and turned it into $38 million of Sequoia's money. The technology is impressive. The discipline is the lesson — and that part is free.