{"slug":"cowork-anthropic","label":"Cowork Anthropic","item_count":3,"day_count":3,"source_count":2,"first_seen":"2026-07-08T00:00:00+00:00","last_updated":"2026-07-22T12:25:00+00:00","generated_at":"2026-07-28T20:08:42.819916+00:00","sources":["claude_blog","infoq_ai_ml"],"days":[{"date":"2026-07-08","items":[{"title":"Using Claude Cowork in marketing operations to automate reporting and campaign building | Claude by Anthropic","url":"https://claude.com/blog/how-anthropics-marketing-operations-team-uses-claude-cowork-to-automate-reporting-and-campaign-builds","source":"claude_blog","type":"news","summary_1line":"How the marketing operations team at Anthropic uses Claude Cowork to automate marketing reporting and marketing campaign builds.","sid":"1eb4b0cca9e92de7","published":"2026-07-08T00:00:00+00:00","editor_note":"Anthropic's marketing operations team details automating reporting and campaign builds with Cowork — the first internal use-case writeup."}]},{"date":"2026-07-16","items":[{"title":"Working with Claude Fable 5 in Claude Cowork | Claude by Anthropic","url":"https://claude.com/blog/working-with-claude-fable-5-in-claude-cowork","source":"claude_blog","type":"news","summary_1line":"Anthropic's Claude Fable 5 can carry long, complex work on its own. Here's how to get the most out of it in Claude Cowork.","sid":"2bef8d685677b14e","published":"2026-07-16T00:00:00+00:00","editor_note":"Guidance on pairing Claude Fable 5 with Cowork for long, complex work."}]},{"date":"2026-07-22","items":[{"title":"Anthropic Details How It Contains Claude Across Web, Code, and Cowork","url":"https://www.infoq.com/news/2026/07/anthropic-claude-containment/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=AI%2C+ML+%26+Data+Engineering","source":"infoq_ai_ml","type":"news","summary_1line":"Anthropic detailed the containment architectures it uses for Claude across its products. It argues that agent safety depends on placing deterministic limits on an agent’s filesystem, network, and execution environment...","why_it_matters":"Matches feed focus: agent.","sid":"86b488f1e921fab7","published":"2026-07-22T12:25:00+00:00","editor_note":"InfoQ reports Anthropic's account of the containment architecture — deterministic filesystem/network/execution limits — behind Claude's agent safety across web, Code, and Cowork."}]}],"editorial":{"tldr":"Anthropic's own teams spent early July showing how they use Claude Cowork — marketing operations automating reporting and campaign builds, then guidance on pairing it with Claude Fable 5 for long, complex work. By Jul 22, Anthropic moved from use-case show-and-tell to disclosing the containment architecture underneath.","stale":false,"whats_new":"Anthropic detailed for the first time how it technically contains Claude across Cowork, Code, and web — arguing agent safety depends on deterministic filesystem, network, and execution limits, not model judgment alone.","why_it_matters":"Cowork is Anthropic's async, hand-off-and-walk-away agent product — this containment-architecture disclosure is the first concrete answer to the platform-engineer question of what actually stops an unsupervised agent from doing damage while you're away.","take_for_builders":"Before handing Cowork an unattended, multi-step task, check what filesystem, network, and execution limits Anthropic actually enforces by default versus what you need to configure yourself.","status":{"state":"Expanding · usage patterns, use cases, and containment architecture disclosed","tone":"rising","changed":"2026-07-22","detail":"A week of use-case posts (marketing-ops automation, pairing with Claude Fable 5) preceded Jul 22's first detailed look at the containment architecture — filesystem, network, and execution limits — that Anthropic says agent safety depends on. No further update since.","track":[{"label":"use cases","detail":"Jul 8-16","tone":"rising","weight":45},{"label":"containment architecture","detail":"Jul 22","tone":"now","weight":55}]},"beats":[{"kicker":"USE CASES","tone":"rising","headline":"Anthropic's own teams show how they use Cowork","summary":"Anthropic's marketing operations team details automating reporting and campaign builds with Cowork, then Anthropic publishes guidance on pairing it with Claude Fable 5 for long, complex work.","sids":["1eb4b0cca9e92de7","2bef8d685677b14e"]},{"kicker":"NOW","tone":"now","headline":"Anthropic details the containment architecture behind Claude's agent safety","summary":"InfoQ reports Anthropic's account of how it places deterministic limits on Claude's filesystem, network, and execution environment across web, Code, and Cowork — a technical answer to what actually contains an unsupervised agent.","sids":["86b488f1e921fab7"]}],"open_questions":["What specific containment mechanisms — sandboxing, permission scopes, network egress rules — does Anthropic enforce by default in Cowork, versus what's opt-in?","Does the containment architecture change how much unattended, multi-step work a team can safely hand to Cowork, or is it primarily a trust-building message?","Do Claude Code and the web app share the same containment primitives Anthropic describes for Cowork, or does each surface implement its own limits?"],"provenance":{"86b488f1e921fab7":{"status_update":true}},"generated_at":"2026-07-28T00:10:00Z"}}