#Agent

共 109 篇带此标签

109 articles Newest 2026-11-19
Archive · #Agent 11 articles
Tools

Inside Socratic Tutor: AI That Questions Before Answering

Jane Street's open-source Socratic Tutor uses follow-up questions to pinpoint gaps in a student's reasoning. Its Agent Skill design shows why effective tutoring depends on strategy switching, state management, and evaluation—not merely prompting.

2026-11-13 Source 2026-11-10 8 min 51 views
Tools

How Rivet Uses Actors to Cut Agent Sandbox Costs

Rivet persists Agent state with S3-backed Actors and hibernates idle instances to reduce resource consumption. Its architecture is best suited to long-running, waiting-heavy workflows, but teams must still validate recovery, concurrency, security, and operating costs.

2026-11-06 Source 2026-11-05 7 min 51 views
Opinion

Claude Duet and the Real Work of Multi-Agent Systems

Claude Duet's significance lies not in a new model, but in how it assigns work across specialist Agents. Its architecture highlights the practical challenges of permissions, shared state, and recovering from failed tasks.

2026-10-30 Source 2026-10-29 7 min 51 views
Case Studies

Game Tests Reveal AI Agent Swarm Engineering Challenges

Y Combinator Labs has deployed AI Agent swarms in Minecraft and Mindustry, exposing issues that game performance alone cannot capture. The experiments highlight duplicated work, state conflicts, coordination overhead, and the metrics needed to evaluate whether multi-Agent collaboration actually works.

2026-10-30 Source 2026-10-29 6 min 51 views
Products

Claude Duet Puts Multi-Agent Collaboration in Plain Sight

Anthropic's Claude Duet is designed around task allocation, information handoffs, and user-controlled review rather than simply adding another chat window. Its real value will depend on whether users can see, influence, and verify how collaborating Agents work.

2026-10-23 Source 2026-10-21 7 min 51 views
Case Studies

OpenAI's Navier-Stokes Proof Failed at Translation, Not Reasoning

OpenAI's attempt to formally verify a Navier-Stokes existence proof exposed a subtle failure mode: the LLM's mathematical reasoning was sound, but the math-to-Lean translation step silently changed the meaning of the proof, passing typecheck while proving a different statement.

2026-10-10 Source 2026-10-09 7 min 51 views