From spinach stems to a trillion dollars — Hermine Tranie does not stop

At nineteen, Hermine Tranie walked out of Imperial College London — one of the world's best universities, where she was studying pure mathematics — because she decided she wanted to be the best chef in the world.

It was not, in retrospect, the last time she would choose the hardest possible version of what she could be doing.

Because training as a chef is no joke. On her first day at one of London's most celebrated Michelin-starred kitchens, she was handed eight kilograms of baby spinach leaves and told to pick the stems. She had no idea how long it would take. Fifteen hours later, she was done.

Turns out, it was all just hazing. The chef walked over, looked at the pile she'd produced, picked it up, and put it in a blender, making the whole stalk-removing process pointless.

"I hated it," she says. "But I also loved it."

That one story tells you something about Hermine, the 25-year-old co-founder and CEO of ClarityCare, the AI company automating regulated administrative decisions in U.S. healthcare.

Today, ClarityCare's autonomous AI operators are powering organizations serving hundreds of thousands of patients across fifty states, and the company crossed a million dollars in revenue within four months of commercial launch.

But the path that got Hermine here is anything but straight.

Nunnery to MIT

Hermine grew up in Paris, a student at a Catholic nun school who was, by all accounts, bored. Not disruptive — just somewhere else in her head. The nuns eventually told her mother as much: she didn't fit their mould, they said. She needed something different.

What she got was Northbourne Park School in Kent, a boarding school that looks, she laughs, "like Harry Potter — and I'm called Hermine, so everyone called me Potter." She was eleven years old. She ran national cross country championships. She learned programming. She was made house captain.

"I discovered a whole new world," she says. "In France, all that mattered was maths. Here there were so many more things to do."

She went back to France for the Jesuits — a more academically rigorous environment, where she was less bored — and eventually made it to Imperial College London to study pure mathematics. Then, after a year, she decided she wanted to be a chef.

Eight kilos of spinach

She isn't joking about the spinach. She put her Imperial degree on pause, prepared for her chef's exam, opened Gloria restaurant in Shoreditch as one of a four-person founding team, and then secured a position at Hide, the Michelin-starred restaurant at Hyde Park Corner. The hours were brutal. The hierarchy was unforgiving. The work was relentless.

"I was the lowest ranking person in the kitchen who still got to cook," she says. The spinach incident was her first day. She wasn't told it was a test. She just did it.

She passed her chef's exam — a six-hour ordeal where anything that could go wrong did. She had never cut herself in a kitchen before. That day, she nearly took off her finger. Standing there, bleeding, wrapping her hand in one of the tiny latex finger covers they keep in professional kitchens. But she passed anyway.

In the end, chef’s life was not for her. She found more meaning in the business building part of opening the restaurant than the cooking. So she decided to go back to Imperial to finish her degree.

She would go on to complete a master's in computer science, talk her way into a coveted spot at École Polytechnique in Paris (while making clear on day one she intended to leave for MIT), and eventually land at MIT itself, where she spent a year researching the extraction and structuring of information from medical records.

After that came Entrepreneur First, where she met her co-founder Alexandru Andrei, who is a 2x founder who had previoudly spent 5 years leading AI initiatives in intake and claims processing for various insurance types (life, title and re-insurance). He’s also an open-source contributor to healthcare standards (HL7, FHIR) and security protocols (ISO, IEEE) still in use today. Together the two started ClarityCare.

Building Decision Infrastructure for Healthcare

Healthcare administration looks complex. Underneath, it's surprisingly deterministic. Prior authorisations, appeals, payment reviews — these are regulated decisions with defined criteria and binary outcomes. The complexity isn't in the decision itself. It's in the chaos of the data you have to wade through to get there: EMRs, portals, faxes, internal notes, all of them fragmented, many of them conflicting.

Most AI companies in healthcare have built copilots that help humans wade through that chaos faster. But the human is still there, which means the cost structure doesn't change. The headcount doesn't change.

Hermine and Alex's insight was simpler and more radical: if the decisions are deterministic, why is there a human in the room at all? ClarityCare's AI operators log into customer systems just like an employee would, reconcile the conflicting records, apply the criteria, make the production decision, and log the audit trail. They go live in three days. Zero engineering work required from the customer.

The thing that makes it possible — and what Hermine believes most of her competitors have missed — is knowing when to stop. "You need to be able to say 'I don't know' when you don't know," she says. "Otherwise there's no point in having an autonomous system." That triage capability, the ability to recognise the edge of its own confidence and hand off a genuine edge case to a human, is what lets customers trust the system with production decisions in the first place. It's the thing that turns a copilot into an operator.

Betting on the unglamorous

It’s not easy to crack the world of US insurance. It took building real trust, industry by industry, relationship by relationship — being at every conference, building a network of internal advocates, proving, case by case, that a machine could be trusted to make a decision without a human in the room. The first time they tried to get a dataset from a customer, it took six months to get a hundred data points. Now, she says, access comes overnight.

ClarityCare is currently processing tens of thousands of prior authorisation, with large government contracts in the pipeline. The business model is per-member: At current pricing and utilisation levels, the path to significant scale is straightforward. The decision infrastructure they're building — the layer that sits above the fragmented systems of record and issues authoritative, auditable, compounding decisions — cannot easily be ripped out. You either replace it with headcount or you find another AI.

At 25, she is building infrastructure for a trillion dollar market that barely knew AI was coming. The nuns in Paris said she didn't fit their mould. They weren't wrong.