
Product Talk How WHOOP turned wearable data into an AI coach | WHOOP AI Product Lead
12 snips
Oct 12, 2026 Viviano Cantu, WHOOP’s AI Product Lead and former software engineer, shares how pizza-fueled experiments became WHOOP Coach. He explores building an AI that understands years of personal physiology, testing it with synthetic personas, and empowering experts through AI Studio. He also discusses privacy, member-controlled memory, clinical AI, emerging agent-builder roles, and bringing engineering rigor to product leadership.
AI Snips
Chapters
Transcript
Episode notes
Why WHOOP Delayed Its AI Launch
- WHOOP delayed its July launch after GPT-4 dramatically improved the prototype’s capabilities.
- The team used the date to maintain momentum while preserving the option to pursue a stronger product.
A Health Bot Is Not Yet A Coach
- A weekend chatbot can produce interesting answers, but a coaching product must reason consistently about one person across years.
- WHOOP invests in evals and internal tooling to scale physiology expertise reliably.
Evaluate AI Coaching Across Multiple Dimensions
- Build evals around measurable dimensions rather than searching for one perfect coaching answer.
- Test whether responses use the right data, avoid excess, fit constraints, and improve engagement or reduce refusals.
