Anthropic Puts Claude’s Whole Office in One Chat
The office suite is now a chat window with ambitions.
Anthropic is merging Claude chat and Cowork into a single interface that decides whether each request needs a quick answer or a longer agentic workflow. The “one Claude” setup also brings Artifacts, skills, connectors, and Claude Design into any conversation, so users no longer have to pick the right AI cubicle before starting work.
New beta tools Claude Docs and Claude Slides can generate editable, shareable documents and presentations inside chats. Users can revise the work manually or by prompting Claude, collaborate through comments, and export to Word, Google Docs, PowerPoint, or PDF. A report and its matching slide deck can come from the same thread—which should at least keep slide 7 from contradicting page 3.
Claude can now keep working on longer tasks while you are offline, ask for approval before taking actions by default, and run recurring scheduled work. Pro and Max users get first access over the coming weeks across web, desktop, and mobile; Team and Free follow later. Enterprise admins choose when to enable the beta tools and will get at least 30 days’ notice before the interface changes.
This is less UI cleanup than Anthropic’s bid to make Claude the place where office work begins and ends. That moves Claude closer to Google’s productivity turf—and asks workers to trust one AI not just to draft the memo but also to choose the tools and actions required.
Google Quietly Adds Claude to Gemini’s Home Turf
Nothing says confidence like inviting your rival to pair-program.
Google has granted its entire engineering workforce access to Anthropic’s Claude Opus 5 for internal programming tasks via its Antigravity platform, though Gemini remains its primary model.
The permission slip comes with caveats. Staffers cannot freely use the standalone Claude Code app, so they must run Opus 5 directly through Google’s proprietary environment, subject to individual usage limits and restricted to specialized jobs.
That is still a sharp reversal. Previously, Google blocked most personnel from utilizing rival programming assistants, carving out exceptions only for specific DeepMind squads and critical initiatives. This broader deployment follows widespread internal complaints regarding Gemini’s capabilities—a tension exacerbated by corporate mandates for staff to increase their output using AI.
The family tree is tangled: Google competes with Anthropic but also plans to invest up to $40 billion in the startup. By funneling Claude through Antigravity, Google can conduct real-world evaluations of a competitor’s model without ceding control over its internal security, quotas, or oversight.
Gemini is still employee of the month. Claude just got a visitor badge.
ChatGPT Tests Ads That Talk Back
OpenAI is testing Sponsored Agents that let users open clearly labeled chats with advertiser-backed AI reps from ChatGPT ads, starting with select US advertisers.
The sales bots can answer follow-up questions about products before sending shoppers to the advertiser’s website. OpenAI says those conversations remain separate from ChatGPT’s independent answers and the user’s original chat.
Meanwhile, advertisers can build, update, and analyze campaigns through natural-language prompts. Ads Manager can also generate copy and images, rewrite text for a conversation’s context, translate ads, and connect with HubSpot and Shopify.
That could compress campaign production while turning ads from passive pitches into interactive sales funnels. It also raises the stakes for keeping paid persuasion visibly separate from trusted AI guidance.
The chatbot is now working the showroom floor.