Samaya AI lead Ashwin Paranjape says earnings agents beat Wall Street consensus on revenue when given live, point-in-time ...

NVIDIA expands Korea sovereign pact to NVLink Fusion with Rebellions and SK Telecom, plus Dynamo work for local models.

Matthew Jewkes argues full automation beats agents, but his 99.95% Clarivate case required 1,000 evals and costly org change.

AlixPartners tested Claude Code on 40,000 contracts. It missed $800K and a renewal date, showing why structure beats prompts.

Tulsee Doshi told Lenny's Podcast PMs matter because models are unpredictable. The frontier claim still lacks shipped proof.

Stanford's Maty Bohacek showed AI agents faking humans to pass phone and Zoom checks, with safety alignment cutting success to 44%.

Uber engineers say AI diffs pushed SubmitQueue to 130 req/hour. Gating and relaxed conflicts hit 300 lands/hour, with broken trunk risk.

AMA co-founder Alfredo told Y Combinator cancer is a data problem and AI agents can build a personal lab for each patient, a ...

Tanmay Sah and Dolly Sah demoed UndoBench and EvoUndo: 84% task success but 47% recovery, with a verifier gate lifting safe ...

Range planner Matt Ragone says 20 planners with no code built 17 live skills, all gated by human review. Matt Ragone, a ...

GitHub's Jay Parikh details secret blocking, Copilot harness and Fusion model routing for 225M developers and agent-created ...

McKinsey on AI Engineer: top 20-25% see 2x software gains by shifting to Kanban loops, small pods and curated context.