Citadel Securities Joao Wenzel told AI Engineer why RAG fails on strategy and how agentic memory tracks evidence decay.

Stephen Chin and Neeraj Deshmukh say graph context fixes mainframe rewrites, citing 35M lines in 4 days and byte parity tests ...

John Langford says Microsoft ended AI Frontiers by moving it into Microsoft AI, sparking a research vs scaling split.

Cloudflare warns AI agents act as intermediaries that hide users, breaking identity and accountability for sites facing dots ...

Anthropic Engineering says offline dreaming lifted Claude from 13% to 48% on 25 questions. Agents now work across sessions.

Paid in Full Foundation gives five-year grants to hip-hop pioneers like Caz and D.O.C. after decades of lost royalties and ...

OpenAI's Fabbri and Steffmann demoed finance agents in ChatGPT Work and Codex, with SEC data, skills and plugins.

Fieldguide's Anirudh Veeraragavan says audit agents need authorization, replayable runs and versioned evidence.

Amjad Masad tells Paul Graham how Replit automated debugging, runs agents to completion and sells deploy depth to enterprise ...

Vinoo Ganesh demoed Kepler pointer verification for finance AI after $270M flowed to rivals, with lookup, security and 10K ...

CNBC traces Anthropic from 2021 OpenAI spinout to November IPO, with $33B revenue, $900B in compute commitments and record ...

Nori Agentic told AI Engineer that cloud agents cut security master checks from an hour to five minutes as private assets ...