‍Unlocking the mysteries of our company's mythical basis

‍

Usually, when people hear the word “centaur,” the first thing that comes to mind is the half-human, half-horse creature from Greek mythology. Given that association, why are we called Centaur Labs, and what does it have to do with medical data labeling and AI anyway?

‍

The History of the Modern Centaur

‍

Fast forward 2000+ years, and the modern centaur, as we use the term today, is a human and machine hybrid. This idea was relatively common in the 1970s but was popularized in 1997 after chess grandmaster Garry Kasparov lost a chess match to Deep Blue, an IBM supercomputer. After some time, Kasparov saw the potential of human and computer teams. Specifically, he explored how they should best collaborate and integrate. He observed the following:

‍“A weak human + machine + better process was superior to a strong computer alone and, more remarkably, superior to a strong human + machine + inferior process.”

Thus, Kasparov invented “Centaur Chess,” which is a form of chess where teams are composed of humans that work with computer/AI support.

‍

Nowadays, the modern centaur concept applies far beyond just chess. Today, humans and AI work together in a variety of fields, from air traffic control to digital art tools, and these types of interactions will only become more prevalent in the future.

‍

Centaur

‍

Centaur is applying the concept of the modern centaur to one of the most challenging, yet not widely known, problems in healthcare: medical data labeling. As the presence of AI/ML increases in healthcare, so does the need for high-quality training data. At Centaur, we’re harnessing the power of collective intelligence by using a network of medical experts (human) to analyze medical data (e.g., images, videos, audio recordings, etc.) and create high-quality data labels. By leveraging human inputs, we’re helping healthcare leaders and innovators accurately train their AI algorithms (machines).

‍
Thus, the Centaur name and mission were born:

“To build a trusted network of people and machines to annotate the world’s medical data.”

‍

Schedule a demo with Centaur.ai

‍

July 1, 2025

Ground Medical AI in Expert-Labeled Data | Centaur AI

Centaur.AI collaborated with Microsoft Research and the University of Alicante to create PadChest-GR, the first multimodal, bilingual, sentence-level dataset for grounded radiology reporting. This breakthrough enables AI models to justify diagnostic claims with visual references, improving transparency and reliability in medical AI.

March 5, 2026

From Prediction Markets to Production-Ready Data

Centaur.ai CEO Erik Duhaime joined SegMed’s Bites of Innovation to explain how healthcare AI teams can achieve data quality at scale. He discusses collective intelligence, why credentials do not guarantee labeling quality, competitive annotation, dynamic escalation using disagreement, confidence scoring, continuous QC, regulatory datasets, de-identification, and how to use LLMs without blind trust.

March 25, 2026

What AI Medical Device Teams Get Wrong About FDA 510(k)

AI medical device teams often underestimate what it takes to achieve FDA 510(k) clearance. Success depends not just on model performance, but on data credibility, expert labeling, study design, and documentation. This post explains how to align with FDA expectations and avoid the common pitfalls that delay or derail submissions.