At a Google Cloud event, Google announced that Gemini is becoming a unified agent for the enterprise. Both sources report that CEO Thomas Kurian described it as an agent that takes 'objectives' rather than 'instructions', plans work, connects to internal systems, and returns finished results. The agent has its own identity and Workspace account, and can create temporary sub-agents for multi-step tasks.
Both sources agree on the model orchestration angle: Gemini picks the best model for each job, with Anthropic's Claude available at launch and other private and open models planned. They also both note new cost controls, including budgets, smart routing, and spend caps. TechCrunch adds that early testers include On, Shopify, and PayPal, and that Google will expand the agent to consumers later.
The sources diverge in tone and detail. The Register emphasizes Google's history of killing products, citing a median lifespan of about 4.1 years across 308 discontinued services, and quotes a Wall Street Journal report that AI spending is nearly impossible to budget. TechCrunch is more promotional, reporting Pichai's claim that nearly 90% of Fortune 100 businesses use Gemini Enterprise and that Gemini has over 1 billion monthly active users. The Register cites a different Pichai statistic: nearly 80% of Google Cloud customers use Google's AI products. The exact numbers differ, but both point to broad enterprise adoption.