How to Leverage Domain Expertise
summary
Claim: embedding domain judgment into the org matters more than squeezing model gains. Presents three operational patterns — Oracle/Evaluator/Architect — with concrete levers: prompt tweaking, review dashboards, hiring clinicians, decentralized prompt variants, and automated improvement pipelines. Gives cases: Granoola keeps a writer-as-oracle for taste on meeting notes (company > billion valuation), Tandem scales with many doctors as decentralized oracles, Anterior evolves from manual review to measured metrics and automated learning. Cautions: 50% of projects drop without these systems.
AI-written summary · about
chapters (10)
- 0:00Leveraging domain expertise
- 2:12Domain expertise in AI
- 4:39Ways to incorporate expertise
- 9:14Choosing the right model
- 11:01Granola’s oracle approach
- 12:49Tandem’s decentralized oracle
- 14:32Anterior’s architecture evolution
- 17:16Skills for domain expert roles
- 19:45Leveraging domain experts
- 23:19Summary
related talks
19mMake your LLM app a Domain Expert: How to Build an Expert System
21mImproving your Developer Experience Through Platforms and Generative AI
48mSpring is bootiful but so is your domain
16marchive – Jason Quinley
14mThe Oracle AI Playbook for Business Leaders
21mCutting through the hype. How enterprises can apply a strategic lens to leveraging AI to achieve ...
38mFrom code to culture: How AI is reshaping the engineering organization
31mExpertise vs Experts