Bio
CVP for Microsoft's AI Platform — infrastructure, foundation models, agent toolchains. Sees the operating constraints of every major AI player. Earlier ran product at Instacart, Porch Group, and WhatsApp. Vocal advocate for the "product as organism" frame and for season-based planning over fixed roadmaps. Predicts post-training and reinforcement learning as the next investment frontier.
Operating themes
- Product as organism. The metabolism (data ingest → reward design → outcome tuning) is the moat.
- Post-training is the new pre-training. Beyond 30B params, fine-tune don't pre-train.
- Seasons, not roadmaps. Plan in capability waves; leave slack for the slope.
- Polymath builders. Hire for the loop, not the lane.
Cards
- The economic moat in AI is post-training on proprietary data, not pre-training a base model — Post-training on proprietary data is the AI moat [Tier A]
- Treat the product as a living organism with a metabolism, not a shipped artifact — Treat product as a living organism with a metabolism [Tier B]
- Plan in seasons keyed to secular changes, not 6-month roadmaps — Plan in seasons keyed to secular changes, not 6-month roadmaps [Tier B]
Sources captured
- 2026-04-28 — Lenny's Podcast, "Product as organism, post-training, agentic society" (
raw/podcasts/asha-sharma--product-as-organism--2026-04-28.md)