Meta AI Gains Task Automation and Daily Briefings on Muse Spark 1.1

Meta Platforms has begun rolling out task automation for its Meta AI assistant, giving the chatbot the ability to carry out recurring jobs without a user re-issuing the same instructions each time. Once someone sets their preferences, the assistant handles the work on its own — a shift from prompt-and-respond conversation toward something closer to a standing arrangement between user and software.

What Meta AI's Task Automation Actually Does

The headline capability is the daily briefing. Meta AI can pull together a summary of what's on a user's calendar and deliver it without being asked. Alongside that, the assistant can build weekly meal plans and push out trend updates on its own schedule.

What ties these three together is the removal of the repeat prompt. Previously, a user wanting a Monday-morning rundown had to request one every Monday morning. Under the new model, the preference is configured once and the assistant executes against it going forward.

Where the Features Are Live

The rollout is limited to select markets at launch, and it arrives through two entry points: the standalone Meta AI app and the meta.ai website. Support inside WhatsApp is on the roadmap but has not shipped, which leaves the largest messaging surface in Meta's portfolio outside the automation features for now.

Muse Spark 1.1: The Model Behind the Automation

The new capabilities run on Muse Spark 1.1, the multimodal reasoning model released by Meta Superintelligence Labs on July 9. That model was purpose-built for agentic work — tool use and multi-step workflows specifically — which is precisely the technical foundation recurring task execution requires.

Muse Spark 1.1 already powers the "Thinking" mode inside the Meta AI app, so the model was doing reasoning-heavy work in the product before automation arrived on top of it. According to Meta's Chief AI Officer, Alexandr Wang, sharpening coding and agentic performance was a central goal of the 1.1 release. The model was also opened to developers in preview form.

Replacing Llama Across Meta's Chat Apps

Muse Spark 1.1 is expected to take over from the Llama models currently driving chatbot experiences across WhatsApp, Instagram, and Facebook. That consolidation matters for the automation story: as the newer model spreads across Meta's messaging surfaces, the technical barrier to bringing task automation into those apps drops accordingly.

Privacy Controls and the "Personal Superintelligence" Framing

Meta positioned the update as movement toward what it calls personal superintelligence — a concept Mark Zuckerberg has described as an AI that understands who you are, what you're aiming at, and what interests you, then helps with whatever you happen to be working on. Task automation fits that description more directly than a conversational assistant does, since acting on stored preferences requires the system to retain a persistent picture of the user.

That persistence is where privacy questions surface, and Meta addressed them on two fronts. The service stays free, and users keep control through customizable settings — preferences drive what the assistant does, so nothing runs without configuration.

Incognito Chat and Private Processing

Meta also pointed back to Incognito Chat, the mode it launched in May. Conversations in that mode run inside a secure environment that Meta says is closed even to the company itself. The feature was designed for sensitive queries and built on the Private Processing technology developed for WhatsApp.

The Capital Behind the Vision

The financial commitment behind this direction is substantial. Meta's capital expenditure for 2026 is projected to reach as much as $135 billion, an outlay that signals the personal superintelligence push is treated internally as infrastructure strategy rather than a feature experiment.

The Race to Ship Autonomous Assistants

Meta is not moving in isolation. The industry is converging on assistants that can act rather than only answer, and the timing here is unusually tight: OpenAI launched ChatGPT Work — with scheduled tasks and computer use capabilities — on the same day Meta released Muse Spark 1.1. Two of the largest players shipped agentic functionality within hours of each other.

The groundwork on Meta's side goes back further. In May, Meta was reported to be testing a highly personalized agentic assistant internally, designed to take on daily tasks for its billions of users. The features rolling out now are the public surfacing of that work.

What Distinguishes Meta's Approach

Two factors set Meta's position apart. The first is distribution: an assistant that eventually runs inside WhatsApp, Instagram, and Facebook reaches a user base measured in billions, without requiring anyone to adopt a new app. The second is price — the service remains free, which removes the subscription barrier that gates comparable capabilities elsewhere.

The unresolved variable is scope. Daily briefings, meal plans, and trend updates are useful but bounded tasks. Whether the same infrastructure extends to more consequential actions on a user's behalf is the question the current rollout doesn't yet answer.