Harpreet Mangat

The Doctor in the Room Where AI Happens

There's a photo in the Getty Images archive. It's 2020. A doctor in full PPE leans over a ventilator screen inside the Royal Free Hospital's infectious disease unit, the UK's front line during the worst of the COVID-19 outbreak. That doctor is Harpreet Mangat - and by that point he'd already been a neurosurgeon, a McKinsey consultant, and a chief of staff to one of the firm's most senior global partners.

He went back to the ICU anyway.

Today, Mangat is Chief Strategy Officer and Chief of Staff to the CEO at Hippocratic AI, the Palo Alto-based company that has raised $439 million and reached a $3.5 billion valuation building AI agents designed specifically for patient communication. The company's headline claim - and the source of its credibility - is clinical safety. Hippocratic's agents do not diagnose. They do not prescribe. What they do, at scale, is engage: answering patient questions, monitoring post-discharge status, guiding medication adherence, and escalating to a human clinician the moment something sounds off.

Mangat is the rare physician at the executive table who can argue the clinical case and the business case in the same breath. His background isn't just decorative - it's structural. His fingerprints are on the safety architecture that makes Hippocratic's AI defensible to hospital systems, payors, and regulators.

Cambridge, Scalpels, and Context-Switching

Mangat grew up in London and trained in medicine at the University of Cambridge, earning a BA, MA, and MD/MBBS with a focus in Medicine and Cognitive Neurosciences - a curriculum that also spanned UCL for global health training. He completed medical electives in Cuba. In 2011, he was selected for the UK's National Neurosurgery Training Programme, one of only a handful of doctors chosen nationally. His specialty: trauma, perioperative care, brain tumors, and a NeuroTrauma fellowship digging into biochemical markers of brain injury.

The operating theatre is a place where you minimize variance. You plan. You eliminate risk. You execute. Mangat was built for it - and by his own account, he was also built for something else entirely. He describes himself as someone who thrives on context-switching: knitting, micro-soldering, fixing cars, playing video games at a near-professional level as a teenager. The same hands that placed electrodes with robotic precision also rewire electronics for fun.

When McKinsey called, Mangat crossed over. He joined the firm's pharma and data science practice before being tapped as Chief of Staff to Dame Vivian Hunt, then Managing Partner of McKinsey UK and Ireland. That role is a different kind of precision - you're running the operating system for one of the most senior executives in global consulting, managing the gap between strategy and execution at pace.

Then COVID-19 hit. Mangat walked away from McKinsey and walked back into a hospital gown. He spent the first wave of the pandemic working as an Intensive Care Doctor at Royal Free Hospital, inside the unit designated as the UK's primary infectious disease response center. His work during that period was documented by Getty Images - a photographic record of a pandemic, and of a surgeon-turned-consultant who couldn't stay away when patients needed doctors.

Sand Hill Road via the ICU

Post-pandemic, Mangat enrolled at Stanford Graduate School of Business, Class of 2022. He arrived as a surgeon who had spent years calculating and eliminating risk. He left - by his own framing - having learned to embrace it. The MBA wasn't a pivot away from medicine. It was an attempt to operate at the intersection of clinical credibility, capital, and technology.

At Stanford, Mangat became a Venture Fellow at Pear VC, the seed-stage firm with a track record in AI, healthcare, and biotech. He subsequently moved into a full venture capital role at Volcano Capital Management, focusing on medical devices. He bootstrapped a startup in genetics infrastructure. He was, by any measure, doing the rounds of Silicon Valley's on-ramps - learning the investment side from the inside.

In June 2023, he joined Hippocratic AI as Chief of Staff. It was a company less than six months old with a bold and, to many, counterintuitive thesis: that the most impactful thing AI could do in healthcare was not be a doctor. It was to be the layer of consistent, empathetic, always-available support that the healthcare system has never had the staffing to provide - and to do it safely enough that doctors and hospitals would actually trust it.

AI That Won't Diagnose You - and Why That's the Point

Hippocratic AI, founded by CEO Munjal Shah, operates on a deliberate constraint: its AI agents cannot diagnose, cannot prescribe, and are always designed to escalate to a human clinician when something sounds clinically concerning. The company calls this the "safety-first" architecture, and Mangat has been central to translating that principle from philosophy into practice.

The company's flagship architecture, called Polaris, is a multi-model LLM constellation designed for real-time patient-AI conversations. The system integrates vocal tone analysis, emotional cue detection, and speech signal processing to understand not just what a patient is saying, but how they're saying it. It escalates when the signals suggest something beyond what an AI agent should handle alone.

Mangat co-authored the Polaris paper (arxiv, 2024) and followed it with a March 2025 study that evaluated 307,000 unique real-world AI calls across healthcare settings - one of the largest empirical validations of a clinical AI system published to date. The study established a new framework called RWE-LLM (Real-World Evaluation of Large Language Models in Healthcare), an attempt to bring the rigor of clinical evidence standards to AI safety validation.

By November 2025, Hippocratic AI closed its Series C - $126 million at a $3.5 billion valuation, led by Avenir Growth with participation from CapitalG (Google), General Catalyst, Andreessen Horowitz, Kleiner Perkins, and Premji Invest. Hospital systems including Universal Health Services, Cincinnati Children's, and WellSpan Health joined as strategic investors. Mangat's title had evolved to Chief Strategy Officer and VP of Strategy - a recognition that the company's growth required the clinical-strategic function he occupies to be formalized at the C-suite level.

The Long Way Around

The Texture of the Person

Mangat is not a one-speed person. He knits. He micro-solders. He fixes cars. He was a professional-level video game player as a teenager. These aren't quirks - they're data points for someone who has always learned by making things with his hands, whether those things are functioning neurons, quarterly strategy decks, or a working circuit board.

The through-line, by his own description, is context-switching. He actively prefers moving across domains rather than tunneling deeper into a single one. That preference has turned out to be an unusual competitive advantage in healthcare AI, a field that desperately needs people who can speak fluent clinician and fluent engineer and fluent operator without getting lost in translation.

Mangat holds an Ethereum identity - mangat.eth - which is either a hedge or an intellectual signal, depending on who's reading. He has described his time at Stanford GSB as a shift from the risk-elimination instinct of surgery to the risk-embracing mindset of startups. He didn't abandon the former - he layered the latter on top of it.

Things Worth Knowing

• Treated Nobel Peace Prize laureate Malala Yousafzai in the ICU following her Taliban shooting - before any of the startup years.
• His ICU work during COVID-19 is preserved in the Getty Images archive - a permanent photographic record of a surgeon who couldn't stay on the sidelines.
• Was a professional-level video game player as a teenager - arguably the earliest version of his systems-thinking career.
• Knits, micro-solders, and repairs cars - the same fine motor skills that made him a brain surgeon make him at home in a machine shop.
• Holds the Ethereum identity mangat.eth, bridging blockchain and healthcare long before that combination stopped sounding strange.
• Helped evaluate over 307,000 real-world AI conversations with patients - that's more clinical validation data than most healthcare AI companies have ever published.