Superpower Launches Its AI Doctor: 140,000 Lines of Code to Replace Your 15-Minute Checkup

TL;DR

Superpower launched its AI Doctor on February 18 - a health companion built on a proprietary LLM trained with input from functional and integrative physicians
The system ingests 100+ biomarkers, symptom history, and wearable data to build a persistent medical profile with lifetime memory
247 commits and 140,000 lines of code across months of engineering, per co-founder Max Marchione
Membership costs $199/year and includes annual comprehensive blood panel at 2,000+ Quest lab locations
The company has raised $34 million across a $30M Series A led by Forerunner Ventures and a $4M pre-seed

"The future is an AI that knows more about your body than any human ever could." That is not a research paper abstract. It is a tweet from Max Marchione, co-founder and CEO of Superpower, announcing the public launch of what his company calls an AI Doctor - the product of 247 commits, 140,000 lines of code, and months of engineering.

The pitch is straightforward: the median physician visit lasts 15.7 minutes. In that window, a doctor must review your history, assess your symptoms, order tests, and prescribe treatment. Superpower's argument is that an AI with perfect memory, 24/7 availability, and access to your complete biomarker history can do the diagnostic reasoning part better - or at least more thoroughly - than a time-constrained human.

The product went live on February 18.

What It Actually Does

Superpower's AI Doctor is a health companion that sits on top of the company's existing biomarker testing platform. The system ingests three categories of data:

Data Source	Details
Biomarkers	100+ markers across 11 health categories (heart, liver, kidney, hormones, metabolic, thyroid, inflammation, nutrients, immune, body composition, DNA)
Symptom history	Every symptom reported, with clinical-level detail: location, quality, onset, aggravating factors, associated symptoms
Lifestyle data	Wearable integrations (Oura, Garmin, CGM devices planned), self-reported diet, exercise, sleep

The core differentiator is persistent memory. When you report a headache, the system records it with the same specificity a physician would during a clinical encounter - except it never forgets. Months later, when you report another headache, the AI correlates the two events, checks them against your biomarker trends, and surfaces patterns like "every time your ferritin dips below 30, your anxiety spikes two weeks later."

Inline Citations and Visible Reasoning

The AI Doctor includes two transparency features that distinguish it from a generic chatbot:

Inline citations: Every claim links directly to the source lab result or previous conversation that supports it. If the AI says your vitamin D is low, you can see which blood draw produced that number.
Visible reasoning: A "Thinking" tab lets you expand the AI's reasoning chain to see how it reached a conclusion, including how it handles missing data.

The Skeptical Companion

The most unusual feature is what Superpower calls "healthy skepticism." The AI tracks consistency between what you say you will do and what you actually do. If you tell it you will cut sugar on Monday and report eating cake on Friday, it will flag the contradiction. If you have a history of stating health goals and abandoning them, the AI factors that into its recommendations - effectively telling you it does not believe you will follow through, based on evidence.

The Technical Stack

Superpower describes its AI Doctor as running on "some of the most capable AI reasoning models available today" - declining to name specific models. The system is built on a proprietary LLM trained with structured input from functional and integrative physicians who review model outputs, score them for accuracy, and retrain weekly.

The knowledge base draws from evidence-based medical literature, functional health frameworks, and longevity research from institutions including Oxford, Harvard, and Stanford. The system is updated continuously as new studies are published.

The engineering scope, per Marchione's announcement: 247 commits and 140,000 lines of code across months of development.

What It Does Not Do

The AI Doctor does not prescribe medication, does not diagnose conditions, and does not replace a physician for acute or emergency care. It operates in the preventive and analytical space - surfacing patterns, recommending lifestyle adjustments, and flagging when professional intervention is needed. Complex cases are escalated to Superpower's human clinical team.

The Business

Superpower has raised $34 million in total funding:

Round	Amount	Lead	Notable Investors
Pre-seed	$4M	Susa Ventures	Day One Ventures, Long Journey Ventures, Atman VC
Series A	$30M	Forerunner Ventures	Winklevoss Capital, Bond Capital, Logan Paul, Steve Aoki, Giannis Antetokounmpo

The membership costs $199/year ($17/month) and includes:

One annual comprehensive blood panel (100+ biomarkers) at 2,000+ Quest lab locations or via at-home collection
Unlimited AI Doctor access
Unlimited concierge messaging with human clinical staff (24-hour response on weekdays)
Personalized health plans with lifestyle, diet, supplement, and Rx recommendations
Biological age calculation and lifetime health tracking

Additional testing (gut microbiome, toxins, cancer screening) is available as add-ons. The service is available in 37+ U.S. states, accepts HSA/FSA payment, and does not bill traditional insurance.

The Clinical Team

The medical advisory board includes:

Dr. Anant Vinjamoori - Chief Longevity Officer (Harvard MD & MBA)
Dr. Leigh Erin Connealy - Founder of The Centre for New Medicine
Dr. Abe Malkin - Founder & Medical Director of Concierge MD
Dr. Robert Lufkin - UCLA Medical Professor, NYT bestselling author

Co-founder Dr. Manoj Arachige left clinical medicine specifically to build the AI Doctor, writing in a public essay: "I realized traditional medicine couldn't achieve large-scale impact because doctors work within broken systems but cannot fix them."

The Traction

Superpower reports over 100,000 members on its waitlist before the AI Doctor launch. The company claims that 63% of members discover early risk factors for diabetes and 70% slow their biological aging speed - numbers that are self-reported and not independently verified.

The celebrity investor list (Logan Paul, Steve Aoki, Vanessa Hudgens, Giannis Antetokounmpo) signals the company's growth strategy: influencer-driven consumer health rather than clinical partnerships or insurance integration.

What It Does Not Tell You

The Regulatory Question

Superpower carefully avoids the word "diagnosis." The AI Doctor "surfaces patterns" and "recommends" - language specifically chosen to stay outside FDA regulatory scope for medical devices. As long as the system does not claim to diagnose, treat, or prevent specific diseases, it operates in the wellness category rather than the clinical one. This is a feature, not a bug - but it means the AI's accuracy is not subject to the same validation standards as an actual diagnostic tool.

The Evidence Gap

The physician-in-the-loop training process - where doctors review and score AI outputs weekly - is a reasonable approach to clinical accuracy. But Superpower has not published validation studies, accuracy benchmarks, or comparison data against standard clinical decision support systems. The 140,000 lines of code are proprietary. The training methodology is described in marketing terms, not research terms.

The Longevity Premium

At $199/year, Superpower is priced for the wellness-conscious consumer who already buys supplements, tracks macros, and wears an Oura ring. The product is explicitly not designed for the people who need healthcare most - those without access, without insurance, without the baseline health literacy to interpret biomarker data. Co-founder Arachige acknowledged this, writing that Superpower will "initially serve the engaged top 10%" before scaling.

Superpower's bet is that the limiting factor in preventive healthcare is not medical knowledge but physician attention. A doctor who sees you for 15 minutes twice a year cannot possibly track every biomarker trend, correlate every symptom pattern, and hold you accountable to every health goal. An AI with perfect memory and unlimited patience theoretically can. Whether that theoretical advantage translates into better health outcomes is a question Superpower will need to answer with data, not commits.

Sources: