Why Google Lets Its AI Rival Into Google Workspace, and What It Means for Your Company
Trending Topics Guest Post
Claude now works inside Google Docs, Sheets and Slides, even beside Gemini. Google’s selling the seat, not the gate — and that may decide who wins AI at work.
Based on reporting by Trending Topics, Guest Post — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Claude has moved straight into Google Workspace. Since 6 October, Anthropic’s assistant can sit in a sidebar next to an open file in Docs, Sheets and Slides, read what’s on screen, and make edits in place after approval. Google’s own Gemini is there too. That setup looks odd at first glance: why invite a rival into your own office suite?
The answer is boring, and that’s why it matters. Google makes money from Workspace subscriptions per user, per month. It doesn’t collect extra just because someone uses Gemini instead of Claude. So blocking Claude would not bring much upside, and it could send Claude-friendly customers toward Microsoft 365. Google’s compromise is to let Claude in as a Marketplace app, which administrators can remove.
Ting Wasner-Lian argues this is really a question about where a platform can charge. She points to DingTalk in China, where Alibaba’s workplace app opened itself to outside foundation models in 2024 and integrated direct competitors such as MiniMax, Moonshot, Zhipu AI, 01.AI and Baichuan AI. The logic there is blunt: keep the enterprise workflow, even if the best model changes. If a platform refuses that choice, the customer may simply leave.
SAP shows the harder version of the same problem in Europe. It has deep control over business data, is pushing its own assistant Joule and its own data layer, and is now facing a case in California from Munich-based Celonis over whether customers should be allowed to share their own data with third-party tools without high fees. That is the real fault line: not model quality, but who controls the gate to the data.
For European companies, the practical advice is simple. Get agent access in writing. Don’t assume normal API access means an AI agent can read and write data the way you need. And map where the critical data actually lives, because that map is the strategy. The next contracts will decide whether your AI stack is flexible — or just rented from a toll collector.
My take — AI-written commentary, not fact-checked reporting
This is the part of enterprise AI everyone keeps pretending is technical when it’s really contractual. The model is often the least interesting piece; the fee at the gate is where the fight starts. Open in principle, pay at the door — that’s becoming the standard business model, and the paperwork will matter more than the demo.
Read more about this at: Trending Topics