Back in January, the viral sensation OpenClaw kicked off the personal AI assistant craze, where you install a team of AI agents on your desktop, give them goals, and set them loose. Fast forward to May and now there’s a new buzzy trend that’s ready to take off: “proactive” AI (i.e., AI agents that go to work without even being asked).
ChatGPT jumped onto the proactive AI bandwagon early, releasing the ChatGPT Pulse feature way back in September 2025. Still in an early preview phase and available only to ChatGPT Pro users on mobile, ChatGPT Pulse delivers a personalized morning briefing based on the topics you’ve recently broached with the chatbot, including data from your connected Gmail inbox and Google Calendar as well as “direct feedback” from you.
Now it appears Claude and Gemini are heading down the same path. Evidence found in the Claude desktop and mobile apps points to a potential new feature called “Orbit,” which is a “proactive assistant” that connects to tools like Gmail, Slack, GitHub, Google Calendar, Drive, and Figma, reports TestingCatalog.
The unannounced Orbit feature, which could be unveiled as soon as Wednesday at Anthropic’s Code with Claude conference in San Francisco, appears to be aimed primarily at developers looking to manage their production workflows, allowing Claude to proactively serve up daily briefings that it researched the night before.
Meanwhile, Google appears to be busy at work on a “Proactive Assistance” feature for Gemini, with Android Authority spotting traces of the feature in the Google app.
Based on screenshots unearthed by Android Authority, Proactive Assistance would piggyback on Gemini’s Personal Intelligence functionality, taking data from your connected Google services (like Gmail, Google Calendar, and Google Drive), your text messages and contacts, what’s on your screen, and your notifications to deliver “personalized suggestions at the right time.”
As with Claude’s rumored Orbit feature, Proactive Assistant for Gemini has yet to be officially confirmed, although Google could unveil it next week at its Google I/O keynote on May 15th.
I’ve reached out to both Anthropic and Google for comment.
All of these “proactive AI” tools would take today’s personal AI assistants to a new level. While agentic AI tools like OpenClaw and Claude Cowork rely on outcomes previously defined by the user, Pulse, Orbit, and Proactive Assistant appear poised to act first, sussing out your interests and intentions based on your chats, connected apps, and other signals.
Would these proactive reports and updates actually be useful? Or simply annoying? That all depends on the implementation. For example, I often find my Claude Cowork automations to be surprisingly helpful (these are automations I asked Claude to set up), while Gemini’s pushy Personal Assistance feature (“That would be perfect for your new Manhattan apartment, Ben!”) has been a major turnoff.
Proactive AI that smoothly anticipates your needs without first needing to be asked is another piece of the personal AI assistant puzzle. Get ready for rising “proactive AI” buzz in the weeks and months ahead.
