AI Summary
→ WHAT IT COVERS Best moments from 2025 AI podcast episodes featuring founders from Harvey, OpenAI, Glean, Abridge discussing breakthrough applications, reasoning models, workforce displacement, and healthcare transformation. → KEY INSIGHTS - **Early GPT-3 Legal Testing:** Harvey tested GPT-3 on 100 landlord-tenant questions with chain-of-thought prompts. Three attorneys approved 86 answers as sendable without edits, revealing legal AI capability before widespread recognition. - **Test-Time Compute Efficiency:** OpenAI reasoning models achieve steeper scaling curves when given tool access. Models defer tasks without comparative advantage to specialized tools, allocating compute more efficiently than pure token generation. - **Bad Markets Become Good:** Enterprise search failed pre-SaaS due to data access challenges. SaaS systems with unified APIs and standardized versions enabled turnkey search products, transforming a graveyard market into viable opportunity. → NOTABLE MOMENT A doctor told her son at dinner that Abridge lets mommy come home early now, explaining she can eat with family every night instead of working late. 💼 SPONSORS None detected 🏷️ AI Applications, Reasoning Models, Enterprise AI