Google Launches Universal Gemini Business Agent Across All Your Apps

Google Launches a Universal Gemini Agent for Businesses

At its "Gemini at Work 2026" event this week, Google unveiled a unified new AI agent that turns Gemini from a conversational assistant into a genuine digital coworker. Instead of juggling separate tools for different tasks, employees can now hand the agent full objectives - not just individual instructions - and it plans the work itself, uses custom skills and tools, and connects to the organization's internal systems to get things done.

The agent gets its own workplace identity, including an email address (@agents.company.com), a calendar, and Drive storage - just like a new hire. Employees can reach it by tagging it, emailing it, or adding it to a group chat, and it maintains the same context across every device: mobile, desktop, and workplace tools like Slack and Microsoft 365.

On the technology side, the agent is also flexible about which model powers it: by default it uses Google's own Gemini models, but it already supports Anthropic's Claude models today, with more models promised in the future. Google will include the agent at no extra cost within existing Gemini Enterprise subscriptions - significantly lowering the barrier for businesses already using the service (Google says roughly 90% of Fortune 100 companies already do).

The announcement lands amid an intensifying race among every major AI vendor for the 'central interface' of enterprise work: Microsoft updated Copilot in late September with a persistent Autopilot agent, OpenAI launched Dots in early October, and Anthropic integrated Claude directly into Google Docs/Sheets/Slides just two days before Google's event. For small and medium businesses, the takeaway is clear: over the coming year, growing access to AI assistants capable of managing entire tasks in the background - not just answering questions - is on the way.

The agent is currently in private preview for select customers, and Google has not yet published full pricing details for broader availability.

Source: TechCrunch | Source: The Verge | Source: VentureBeat

Microsoft's CEO: "Assume Every AI Model Is Already Compromised"

In a lengthy post on X, Microsoft CEO Satya Nadella called for building much stricter control mechanisms around advanced AI models. He argued we can no longer treat AI as "nested black boxes" whose advice and actions we simply accept or reject. He's calling for a more transparent system where a model's actions can be contained, observed, and documented in tamper-proof, human-readable form.

His most notable recommendation: "We must assume a model is compromised and contain it from the start. Think of it like an emergency brake - an authorized person should always be able to pause or shut down a model mid-task." This call joins a broader industry trend toward incident transparency, independent audits, and data verification - underscoring that as businesses expand their use of autonomous AI agents, human oversight mechanisms need to keep pace.

Source: The Verge