Google outlined an aggressive expansion of its AI portfolio at its I/O developer conference, centering on “agentic” capabilities that perform tasks on users’ behalf. The company introduced Gemini Spark, a cloud-based assistant designed to proactively handle routine work—from parsing meeting notes and emails to drafting action items—while seeking permission for higher‑stakes actions. Google said its Gemini app’s monthly active users have climbed to more than 900 million.
The latest model family, Gemini 3.5, begins rolling out with the speed‑focused 3.5 Flash, now the default in the Gemini app and Google Search’s AI mode, with a 3.5 Pro to follow. Google also debuted Gemini Omni for multimodal video generation and editing, with imperceptible SynthID watermarking and expanded content‑credentials checks; Omni Flash is available to paid AI tiers and, at no cost, on YouTube Shorts and the YouTube Create app.
Search gains an “intelligent” box for longer, multimodal queries, while a Universal Cart will track prices and restocks across services. Alphabet has poured billions into AI infrastructure, with this year’s capex potentially reaching $190 billion, as shares have risen since its latest earnings beat.
Related articles:
– OpenAI unveils GPT-4o, a real-time multimodal AI model
– Microsoft introduces Copilot+ PCs and new on-device AI features
– Meta announces Llama 3 and expands the Meta AI assistant across apps
