The personal AI assistant phenomenon began gaining traction in January with the popular OpenClaw software, which lets users deploy groups of AI agents on their computers, assign objectives, and let them operate independently. By May, attention has turned to an emerging movement: proactive artificial intelligence, meaning agents that initiate tasks without explicit instructions.
OpenAI's ChatGPT was an early adopter of this approach, introducing the ChatGPT Pulse capability in September 2025. Currently in initial testing and limited to ChatGPT Pro subscribers via mobile devices, this tool provides a customized daily summary drawn from recent conversations with the AI, integrated information from linked Gmail accounts and Google Calendars, plus user-provided input.
Recent discoveries in the Claude applications for desktop and mobile suggest that Anthropic is pursuing a similar direction. Traces of an upcoming tool named 'Orbit' describe it as an initiative-driven helper that integrates with services such as Gmail, Slack, GitHub, Google Calendar, Drive, and Figma, according to insights from TestingCatalog.
This undisclosed Orbit functionality, potentially debuting as early as Wednesday during Anthropic's Code with Claude event in San Francisco, seems targeted at software developers seeking to streamline their operational processes. It would enable Claude to generate overnight-researched daily overviews proactively.
On the Google side, development traces of a 'Proactive Assistance' option for Gemini have surfaced in the Google app, as noted by Android Authority.
From images revealed by Android Authority, this Proactive Assistance would build upon Gemini's Personal Intelligence system, incorporating details from linked Google accounts including Gmail, Calendar, and Drive, along with text messages, contact lists, on-screen content, and alert notifications to offer timely, tailored recommendations.
Similar to the speculated Orbit for Claude, Gemini's Proactive Assistance remains unverified officially, though Google might announce it the following week at the Google I/O presentation on May 15th.
Contact has been made with Anthropic and Google for official statements.
Such proactive AI capabilities represent an advancement for current personal assistants. In contrast to workflow-oriented systems like OpenClaw and Claude Cowork, which depend on user-specified results, features like Pulse, Orbit, and Proactive Assistance seem designed to anticipate user preferences and goals through conversation history, app integrations, and various indicators.
The practical value of these unsolicited reports and alerts hinges on their execution. For instance, the author frequently appreciates the effective automations created via Claude Cowork at the user's request, whereas Gemini's intrusive Personal Assistance suggestions, such as commenting on apartment suitability, have proven irritating.
Seamless proactive AI that predicts requirements without prompting fits into the broader evolution of personal AI companions. Expect increased discussion around proactive AI in the coming period.
With over two decades of coverage in consumer tech journalism, Ben now concentrates on artificial intelligence's impact on everyday human activities. His reporting examines recent large language models and their applications in professional and domestic settings to help navigate the ongoing AI transformation. 'AI will reshape our existence more rapidly than anticipated,' Ben observes. 'The optimal adaptation strategy involves incorporating it into daily routines.' A contributor to PCWorld since 2014, Ben has reported on topics ranging from portable computers to surveillance devices prior to establishing the publication's AI focus. His work has also featured in PC Magazine, TIME, Wired, CNET, Men's Fitness, Mobile Magazine, and other outlets. Ben possesses a master's in English literature.