Meta AI announced a package of feature updates on Friday that give the assistant the ability to formulate plans, connect to users' email and calendar applications, and carry out tasks from start to finish. The new functionality is supported by Muse Spark 1.1, Meta's AI model introduced earlier this month.
With these updates, the assistant can now generate multi-step plans and progress through subsequent actions without repeated prompting. Meta highlighted practical uses to illustrate the capability: the assistant can learn a user's aesthetic preferences, search Marketplace for furniture that fits a budget, and compile a mood board to assist with a kitchen renovation. In another example, it can construct a week-by-week training schedule for a half marathon and adapt that regimen to a user's availability.
Meta AI also added a calendar-aware daily briefing feature. The briefing draws information from connected calendars, flags scheduling conflicts or changes, and delivers a concise summary at a time chosen by the user. The assistant supports recurring tasks as well; once a task is scheduled on a repeated cadence - such as weekly meal planning or regular topic updates - Meta AI will continue to deliver that content without repeated setup.
On the research and content front, Meta AI can gather information from across the web, including research papers and material shared within Meta's own apps, and use those findings to produce slide decks. Users are able to provide feedback while the assistant prepares reports, presentations, or plans, and all generated content is consolidated in a single location where it can be reviewed, edited, and shared.
The new capabilities began rolling out on Friday in select markets via the Meta AI app and at meta.ai. Meta said it plans to extend availability to more countries and platforms in the coming weeks, including WhatsApp.
Clear summary
Meta AI's latest update, powered by Muse Spark 1.1, introduces automated planning, calendar and email integration, recurring task delivery, web-scale research aggregation, and content generation, with an initial phased rollout and planned expansion to additional platforms including WhatsApp.
Key points
- Meta AI can create and execute multi-step plans without repeated user prompts, exemplified by home renovation and half-marathon training scenarios.
- The assistant now integrates with calendar and email apps to provide daily briefings, detect schedule conflicts, and manage recurring tasks.
- Research aggregation and slide generation are supported, and all output is stored centrally for user review and modification.
Risks and uncertainties
- Phased availability - features began rolling out in select markets, creating uncertainty about when specific users or regions will gain access; this impacts technology and communications services.
- Dependence on a single model - the new capabilities rely on Muse Spark 1.1, so performance and limitations of that model will directly affect feature behavior and reliability.
- Feature refinement - user feedback is part of the workflow for improving reports and presentations, indicating that initial iterations may require user input to reach desired quality levels.