AI Pulse · AI Trends Pulse

Google just took Gemini Spark, the background AI assistant that runs constantly to manage your email, calendar, and documents, from a US-only test to more than 160 countries. If you pay $20 a month for AI Pro, you can now turn it on almost anywhere. The big exception is Europe (the EEA, UK, and Switzerland) plus Nigeria, probably because of local data and privacy rules that Google hasn’t sorted yet.

The new piece is Chrome auto-browse. Spark can now open your browser, log into sites you’ve saved credentials for, and book flights or restaurant reservations on your behalf. It’s not just reading your inbox and suggesting replies anymore. It’s acting, which crosses a threshold most business owners haven’t thought through yet.

What this means for you

If you run a company, the question isn’t whether an AI can book a table. It’s whether you want a background process with saved logins touching customer systems, vendor portals, or internal tools while you’re asleep. The convenience is real, but so is the risk surface. One misconfigured permission or one hallucinated action and you’ve got a mess that’s hard to audit after the fact.

This is exactly the kind of capability we wire into an AI command centre with guardrails, not as a free-roaming agent. You want logging, approval gates, and scoped access, especially when the AI is holding keys to systems that matter. Spark is impressive, but it’s built for consumer convenience, not enterprise control. If you’re thinking about deploying something similar inside your business, start with the control layer first. The automation is easy. The governance is what keeps you out of trouble, and according to the original report, Google’s rollout shows they’re still navigating that line themselves.