From Academia to Building Thymia - Emilia Molimpakis

Why It Matters

Themeia turns everyday voice interactions into a reliable health‑monitoring tool, opening new revenue streams for healthcare, automotive safety and AI assistants while addressing a critical need for real‑time, non‑invasive diagnostics.

Key Takeaways

• Themeia uses 15‑second voice clips to detect health states.
• Built world’s largest multimodal voice dataset—100k participants, 160 countries.
• Partnerships span NHS, pharma, and automotive driver‑fatigue monitoring.
• Data collection relied on gamified clinical trials to engage users.
• Emphasis on actionable insights, not just biomarker accuracy.

Summary

In this interview, Emilia Molimpakis explains Themeia, a voice‑biomarker startup that extracts physical and mental health signals from just 15 seconds of speech. The company leverages a proprietary, multimodal dataset—now approaching 100,000 participants across 160 countries—combining voice, video, wearables, blood markers and EEG to train models that detect fatigue, stress, depression, anxiety, respiratory issues and even early signs of diabetes. Key insights include the technical breakthrough of real‑time state detection, the importance of clinically labeled longitudinal data, and the strategic focus on trust‑building partnerships. Themeia has secured pilots with the UK NHS, major pharmaceutical firms, and automotive manufacturers to monitor driver fatigue, demonstrating the technology’s safety‑critical credibility. Molimpakis highlights how the team gamified clinical protocols to recruit and retain participants, turning neuro‑psychology tasks into simple video games. She also shares concrete acoustic cues—flat intonation for depression, friction in breathing for respiratory health, and altered vocal cord control for diabetes—illustrating the depth of the signal extraction. The broader implication is a future where voice agents continuously gauge user state, enabling personalized interventions in healthcare, customer service, and transportation. As AI proliferates, Themeia’s focus on actionable, ethically‑sourced insights positions it to become the de‑facto “spell‑check” for human condition in real time.

Original Description

Alice Bentinck sits down with Emilia Molimpakis, founder of thymia, for a Waymo conversation in San Francisco.

thymia is building voice biomarker technology that can detect signals related to mental health, stress, fatigue, respiratory health, and even diabetes from just 15 seconds of voice.

In this episode, Emilia shares the story behind thymia: moving from academic research to founding a company, meeting her cofounder Stefano at Entrepreneurs First, building AI before the LLM wave, and spending time in San Francisco as a UK founder.

They also discuss why voice is such a powerful signal, how thymia built one of the world’s largest clinical voice datasets, and why customer speed in the Bay Area can change what founders believe is possible.