Every Claude Cowork story we've covered — releases, tutorials, and analysis, summarised from the community and the official changelog. 15 and counting. New summaries are published as videos drop.
Looking for guides instead? See the Claude Cowork hub.
A Google DeepMind engineer walks through Gemini managed agents: the Interactions API call, environment reuse across agents, mounting AGENTS.md, skills and a repository, and credential injection. Checked against Google's docs, which differ from the talk on the selectable models and say nothing on triggers.
The multiagent_20261001 config, session budgets in cents, user.interrupt, run limits and vault credentials, from Anthropic's AI Engineer workshop, checked against the official docs.
The 5pm-to-morning agent loop a CTO uses to ship code around a manager's calendar — context gathering, verification baked into the prompt, and the org scaffolding that makes it safe.
Two Anthropic Applied AI engineers walk through how agent surfaces evolved from the Messages API to the Agent SDK to Claude managed agents, the agent/environment/session primitives, credential vaults, self-hosted sandboxes and MCP tunnels.
Most people use Claude Cowork like a chat window. This video shows the right way: projects, connectors, and skills that let Claude handle 90% of the work autonomously.
Craig Hewitt demos 12 schedulable Claude CoWork skills — morning briefing, inbox triage, meeting prep, brain dump — connected to Gmail, Calendar, and Google Drive for fully automated knowledge work.
Allie K. Miller demos Claude Cowork's live artifacts feature—interactive dashboards that pull real data from Fireflies, Drive, and calendar. Claude auto-detects connectors and builds the dashboard without explicit instructions.
Four new Claude managed agent features: Dreaming improves agent memory between sessions, Outcomes adds self-grading, orchestration runs specialist agents in parallel, and webhooks connect outputs to external tools.
Claude Cowork's Live Artifacts connects Gmail, Calendar, ClickUp, and Airtable into one no-code dashboard. Data refreshes on open; interactive buttons can update tasks in connected tools.
Claude Cowork Live Artifacts connects Claude-built HTML components to real tools like Gmail, Google Sheets, Notion, and Fireflies for live updating dashboards and trackers.
Bart Slodyczka shows how to run Claude Cowork entirely on your own hardware using Ollama as the model backend — zero API costs, full data privacy, no cloud dependency.
Bart Slodyczka's complete Claude Cowork tutorial: building skills, setting up recurring automated tasks, and designing workflows that handle your daily work without human input.
Full walkthrough of Claude Managed Agents: create an agent in the console, test it live, then deploy as a production service using N8N for session management.
Code is more powerful, Cowork is easier. Allie K. Miller's decision framework: use Cowork for external apps, Code for software development and automation.
After 3 hours testing Claude Managed Agents, Nate Herk's verdict: excellent for non-technical users, but lacks native crons and multi-agent orchestration.
← Back to the full news digest · Browse the Claude Cowork guides