Centaur Labs

/ Startups / By

Founders: Erik Duhaime, Tom Gellatly, Zach Rausnitz
Founding: 2017
Mission: Accelerate AI breakthroughs in science and healthcare
Employees: 30 & 50% Local
Workplace: Hybrid
Stage & Capital Raised: Series A & $15.9M Raised
Investors: Accel, Matrix Partners, Omega Venture Partners, Susa Ventures, Y Combinator
Key Customers: Medtronic, Eko, Massachusetts General Hospital, Volastra
Glassdoor Rating: N/A
Valuation (estimated): $25M – $100M (assuming they sold ~20% of the company in the $15M Q3 2021 Series A fundraise)
^ this is a useless number. There is no tangible valuation until the business is sold or goes public. Don’t forget it!

Centaur Labs is the leading healthcare data annotation platform — helping AI developers build, monitor, and validate their models at scale. Centaur leverages a community of 58K+ healthcare experts and performance-based incentives, to drive fast and accurate annotations across multiple data types (text, 2D/3D imaging, video, audio, waveform, etc.). 

Centaur works with prestigious institutions like Memorial Sloan Kettering, Mass General Brigham, Paige.AI, Eko Health, and Consensus. Founded by Erik Duhaime, Tom Gellatly & Zach Rausnitz in 2017, this team is powering scientific breakthroughs by annotating biomedical datasets at unprecedented scale and quality, with a unique collective-intelligence based approach.

What does that even mean? 

Well, Erik got his PhD at MIT’s Center for Collective Intelligence where he studied how groups of people, enhanced by AI and other technology, can collectively be more than the sum of their parts. Simultaneously, his wife was in medical school studying to be a surgeon, and was regularly using  flashcards and online quizzes to study for her exams. Erik thought there may be a way (via collective intelligence) to make her studying useful to more than just her, maybe even..profitable? He teamed up with his Brown college friends Zach and Tom, each accomplished in their own right, to set about making a dent in the healthcare universe and figure out if you could get a group of distributed medical workers to analyze biomedical information as well as a fellowship trained expert doctor.

Did you know 30% of the world’s data volume comes from the healthcare industry (RBC)? And no, that is not expected to slow down. AI developers that want to leverage biomedical data to build AI have a difficult time getting their data cleaned, structured and ML-ready. You need substantial datasets and scalable high quality data labeling pipelines to build and manage effective models. There’s a real problem to be solved in helping AI developers structure their datasets quickly, at a high quality, and an affordable price.

Centaur Labs has built a collective intelligence-powered platform (i.e. network) of 58,000+ qualified labelers who help annotate biomedical data of all types – from text to 3D images – for their clients across the healthcare ecosystem, from medical device leaders like Medtronic, to pharma and household names in big tech. The crowd can often outperform individual experts because assessing a lesion is just one highly-specialized learnable skill, while being a dermatologist represents a wide variety of thousands of skills. This intelligent medical crowd (think medical students in Ghana, radiologists in Vietnam, nurses in Nebraska) competes in labeling contests for prizes on Centaur’s iOS app in the app store. They’ve turned what is typically pay per hour work into free (and fun!) “work”, labeling millions of pieces of data any given week.

Anyone can download the app and sign up, so they are able to collect millions of opinions weekly. The platform has quality control mechanisms built in – they’re continually measuring user performance on hidden ground truth to determine what labelers do a good job, and only top annotators get cash prizes. Think of it as an intelligent audit process to determine which opinions to trust.  

The intelligent crowd platform is their bread and butter. However, being experts in the art of labeling, Centaur Labs also helps facilitate a “bring your own labelers” model where customers can then use their quality control software technology for their own teams. For customers who are submitting their AI models to regulatory agencies, they also offer a custom consulting solution. Centaur will recruit, train and manage labelers who have the unique and rare qualifications needed to label validation datasets. For example, FDA approvals may mandate that labeling is done by a US board certified and fellowship trained neuroradiologist with 10 years of experience. No shortcuts!

Today Centaur Labs has almost 30 employees, top tier funding from Y Combinator, Matrix Partners, Accel, Susa Ventures and is processing 2M labels per week. They’ve processed over 175M labels in total since founding.

2023 was a big year – the company published 7 papers with academic research partners to prove out the methodology, closed their biggest deals to date, and are better differentiated and positioned than ever to capitalize on AI tailwinds in 2024. They released new APIs, a desktop labeling experience powered by Meta’s Segment Anything model (SAM), cleared SOC2 Type 2 and HIPAA audits, and released their new Validation dataset solution. This coming year they’re excited to deepen partnerships in the ecosystem, grow the sales and CS team, and further penetrate healthcare segments where AI investment is growing rapidly. 

Operators to Know (Locally):

Key Roles To Be Hired:

If I were interviewing here are some questions I’d ask:

  • Which of the various business lines are the most strategic to Centaur Labs’ success?
  • What are the biggest challenges to growing the marketplace and adding various customer use cases in the quarters ahead?
  • What is the long term vision for the company? How does the recent LLM explosion change that approach?
  • What are the most important roles the team will be looking to add in 2024 // teams that are positioned for the most growth?

We’re optimizing for readability here so to learn more about Centaur Labs you’ll have to D.Y.O.R. I’m excited to watch this team help bring more medical breakthroughs to humanity. The future is looking downright thrilling. See you around town!

Subscribe
Subscribe