Gemini Spark is now live as Google’s first personal AI agent — and it’s built into Gemini for Android, iOS, and Mac, not a standalone app. It runs in the cloud, works while your device is locked or closed, and handles recurring tasks without manual prompting. No subscription fee. No separate download. It’s available to all Gemini users with access to the latest version of the Gemini app or web interface, including free-tier users. As of October 2026. No price is attached: it’s part of Gemini, not a premium add-on. No launch date was announced in the reporting, but Android Police published its hands-on on October 9, 2026, confirming active use by at least one tester. So it’s shipping now, quietly, globally, and without fanfare. Not a beta. Not a waitlist. Just there. You open Gemini, tap ‘Spark’, and start scheduling. That’s it. The feature doesn’t require Pro, Ultra, or any paid tier. It’s in the base layer. And it’s already changing how people structure their day. Not with flashy demos, but with small, scheduled, repeatable acts of delegation. Like checking email and calendar every weekday at 10 AM. Or scanning AI tool updates each morning. Or summarizing NFL results every Monday. All triggered by plain-language prompts. All executed without you watching. All tied to real apps: Gmail, Calendar, Docs, Sheets, and, on Mac. Local folders. That last bit matters: on Mac only, Spark can read and organize files from your Downloads or Screenshots folder. That capability isn’t confirmed for Android or iOS yet. No word on when it might arrive there. No mention of WhatsApp, Slack, or third-party app integrations beyond Google’s own suite. No mention of voice triggers, offline mode, or cross-device sync beyond what’s implied by cloud execution. What is clear: this isn’t another chatbot. It’s an agent. It remembers goals. It acts over time. It waits for deadlines, not just prompts. And it’s already in use, not as a demo, but as infrastructure. People are using it to stop doing the same thing every morning. That’s the shift. Not faster answers. Fewer things to do at all. Next up: wider rollout to more languages, deeper app permissions, and. Possibly, enterprise controls. But right now, it’s live. Free. And already running in the background of someone’s workday.