Google AI Updates: October 9, 2026

1. Google Cloud Introduces the Gemini Agent at Gemini at Work 2026

Google. At Gemini at Work 2026 on October 8, Google Cloud CEO Thomas Kurian introduced the Gemini agent, a single agent for work with one interface and one API. It runs in the cloud with persistent memory, supports scheduled and event-triggered tasks, and is reachable from web, mobile, desktop, the command line, Workspace, Microsoft 365, and Slack, or can be embedded headless in third-party apps. It spins up temporary sub-agents for multi-step jobs and can create persistent “coworker agents” that get their own Workspace account, email address, and calendar. Model orchestration covers Gemini and Anthropic’s Claude models today. Google also announced Smart Routing, hard per-project spend caps in the Cloud Billing Console, an Agent Gateway firewall with Agent Sandbox isolation, preview Financial Services and Legal specializations, and TPU 8i with a claimed 80% better price-performance than the prior generation. Source

2. Google Research Field Trial Finds AI Raised Lawyers’ Output but Not Junior Expertise

Google. Google Research published results on October 7 from a three-month randomized trial with 133 patent attorneys at eleven law firms, who used an unreleased Google Labs patent drafting assistant that is now part of Gemini Notebook. AI access raised drafting scores by 0.34 standard deviations at 10 days and 0.38 at 90 days, and juniors drafted 18 minutes faster than the control group’s 124-minute average. On a day-90 redlining task done without AI, treated seniors gained 0.45 standard deviations while juniors showed no measurable improvement. The authors, David Autor and Tanya Rodchenko, conclude that the output gains did not become lasting expertise for junior lawyers. The full paper is on NBER. Source