VideaHealth: Building the AI Factory

Florian Hillen, co-founder and CEO of VideaHealth, a startup that used artificial intelligence (AI) to detect dental conditions on x-rays, spent the early years of his company laying the groundwork for an AI factory. A process for quickly building and iterating on new AI products, Hillen believed that the AI factory would prove a competitive advantage when he took his company to market. The AI factory consisted of four stages: the data pipeline, which required labeling dental pathologies on x-rays; algorithm development, which required writing computer programs that learned to recognize the presence of dental conditions and identify them in un-labeled x-rays; an experimentation platform, which tested the accuracy of Videa’s diagnoses and determined the appropriate sensitivity of the AI models; and software infrastructure, which deployed the company’s AI programs and managed the process for cleaning, storing, and transmitting data. Hillen knew that his company’s technology would appeal to individual dentists or dental service organizations (DSOs), who could use AI to make diagnoses more accurate and improve patient trust. Yet Videa would also benefit insurance companies, who could use the technology during the claims review process. Poised to take the company to market, Hillen must decide which customer segment to pursue.