Anthropic brings Memory for Teams (with Incognito chats) to Claude

Anthropic brings Memory for Teams (with Incognito chats) to Claude

Rolling out now to Team & Enterprise—admins can disable; Incognito chat available to all users.

  • What’s new: Claude can remember your team’s projects, processes, client context, and priorities—so you don’t have to re-explain every time. Incognito chats let you work without saving history or updating memory. (Anthropic)

  • Granular controls: Memory is optional with in-product visibility and editing; admins in Enterprise can disable org-wide at any time.

  • Project boundaries: Separate memories per project to keep contexts isolated (e.g., launch plan vs. client work).

Why it matters

  • Fewer handoffs, faster follow-ups: Long-running work (sales cycles, sprint planning, exec initiatives) keeps context across chats.

  • Privacy by default when needed: Use Incognito for sensitive brainstorming or fresh sessions without prior context.

Availability

  • Rolling out today to Team & Enterprise; Incognito is available to all Claude users. (Anthropic)

Other News

  • Google Cloud makes GKE Inference Gateway GA with features aimed at lower TTFT/TPOT, prefix-aware routing, disaggregated serving, and model streaming—part of the AI Hypercomputer stack. (Google Cloud)

  • Gemini CLI adds open-source extensions: /security:analyze scans your repo for vulns; /deploy ships to Cloud Run from the terminal (works via MCP). (Google Cloud)

  • Google Ads: AI Max expands globally (beta in Google Ads, Editor, SA360, API) for one-click AI optimization across Search campaigns. (blog.google)

  • Google Ads – Asset Studio: new generative creative suite to generate/scale assets; PMax & Demand Gen updates called out. (blog.google)

  • GitHub: “Copilot coding agent 101” guide shows task-level, agentic workflows that write, run, and test code—tracking from issue → PR → review. (The GitHub Blog)

  • AWS SageMaker HyperPod: managed tiered checkpointing announced—writes fast checkpoints to CPU RAM with async replication and periodic S3 backup; tested up to 15k+ GPUs. (Amazon Web Services, Inc.)

  • Microsoft (Azure AI): “Agent Factory” update on wiring agents/apps/data via MCP & A2A open standards (thought-leadership, but useful for builders). (Microsoft Azure)

See you on Monday!

AI Need That