Inside Anthropic: How Lesse and Jiang Ship Opus 5.5 and Managed Agents
Latent Space · youtube · 2026-10-28
Latent Space interviews Anthropic's Heads of Engineering and Product, Katelyn Lesse and Angela Jiang, on the systems behind Opus 5.5: the 'brain and the hands' architecture for Claude Managed Agents (sandbox can die without killing the agent, with harness/session state/execution split); credential isolation so agents use credentials without holding them; rewriting the Messages API from Python to Rust with Claude and verifying via shadow traffic; anti-distillation measures; memory, 'dreaming,' and agent evals; plus swyx's breakdown of Anthropic's S-1 (revenue per employee, compute efficiency).
More from coding & agent
- Free Claude Code skill repo adds 50 video effects via one /effects command — aitrendz_xyz · 2026-10-28
- Cursor ships Annotate in design mode for visual agent prompts — dotey · 2026-10-28
- Open-source agent-certify baselines your agent system and alerts on anomalies — CognitiveEngineering · 2026-10-28
- "I want an AI assistant that can Irish exit on my behalf," dev quips on X — vhanagwal · 2026-10-28
- Dev ships MyGo, a Go desktop GUI framework, in three weeks using Opus 5.5 — dotey · 2026-10-28
- Karpathy's GitHub repos as a free LLM learning path: micrograd to nanochat — lmoroney · 2026-10-28