The news
On July 24, 2026, Meta Platforms (META) gave its assistant new Meta AI agent features, saying it can now plan tasks, connect to email and calendar apps and act on users' behalf. The features run on the company's Muse Spark 1.1 model.
Meta described the change as its next step toward what it calls personal superintelligence: an assistant that knows a user's context and handles tasks for them. The company said the features would begin rolling out that day in select markets in the Meta AI app and on meta.ai, with more countries and surfaces, including WhatsApp, to follow in the coming weeks. The post did not name the markets or the email and calendar services supported.
Meta's examples showed multi-step work. Asked to plan a birthday dinner, the assistant looks for restaurants and checks the user's calendar for a free evening. Asked about a first half marathon, it drafts a weekly training schedule and sends the coming week's plan every Monday. A new daily briefing reads the user's calendar, flags double bookings or changed plans, and arrives at a time the user picks.
Users set up a recurring job once, such as a weekly meal plan or alerts about new product drops, and Meta AI keeps delivering it without further prompts, the company said. The assistant can also research a topic across the web and Meta's own apps, turn the findings into slides, and take new instructions mid-task, such as a change of focus or tone. Its output is stored in one place for later editing and sharing.
On privacy, the July 24 post pointed users to Incognito chats for fully private conversations. It did not describe what level of access people grant when they connect an email or calendar account, or how that data is stored.
The numbers
- Model behind the features
- Muse Spark 1.1 (released July 9, 2026)
- Context window
- 1 million tokens (Meta)
- API price, input
- $1.25 per million tokens (TechCrunch, citing Reuters)
- API price, output
- $4.25 per million tokens (TechCrunch, citing Reuters)
Why CEOs should care
For CISOs and IT leaders, the practical change is that a consumer assistant now invites people to connect email and calendar accounts, and many employees use the same accounts for work. Meta's announcement did not spell out access levels or data retention for connected accounts. Security teams should check whether staff can approve third-party app access to corporate mail and calendars on their own (the OAuth consent step, where a user lets an outside app into an account) and tighten that policy if they can.
For buyers of workplace productivity assistants, Meta AI now covers overlapping ground in a consumer app: daily briefings, scheduling help, web research and slide drafts. Expect employees to compare. When renewing a paid assistant, ask the vendor which tasks run unattended, what approval is required before the agent acts, and what audit log administrators can see.
For CFOs and boards, the model's price is the more lasting signal. TechCrunch, citing Reuters, reported on July 9 that Meta would charge developers $1.25 per million input tokens and $4.25 per million output tokens for Muse Spark 1.1, in line with, though slightly above, Anthropic's Claude Haiku 4.5 and OpenAI's GPT-5.6 Luna. Boards should ask how often AI budgets are revisited as per-token prices move.
The bigger picture
The launch followed a busy July for Meta Superintelligence Labs, the unit that built Muse Spark 1.1. Meta released the model on July 9, 2026, describing it as a multimodal reasoning model built for agentic tasks, with a context window of 1 million tokens, and opened it to developers through a Meta Model API in public preview. TechCrunch reported that the first Muse Spark version was announced in April and that the new model aims to compete with similar products from OpenAI and Anthropic.
Meta's bet is that an assistant becomes more useful when it works inside the apps where people keep their schedules and messages. That moves a consumer product toward the territory of workplace tools, where inbox and calendar access is the core of what customers pay for.
What happened next
On September 8, 2026, Meta introduced Muse, a separate personal AI agent that runs on a dedicated cloud virtual machine and can send email, book travel and buy things. Meta said Muse asks before sensitive actions such as sending an email or making a purchase, keeps an audit trail, and lets people choose whether it only reads their email or can also send. It began rolling out in the US on iOS, Android and muse.ai.
On September 23, 2026, at Meta's Connect event, CEO Mark Zuckerberg called Muse the centerpiece of the company's vision, and Meta announced retail partnerships including Best Buy, Walmart and Wayfair, as reported by TechCrunch. Next to watch: Muse Confidential VM, which Meta says will encrypt a user's data with a key only they hold and is due later this year, and Muse on Meta's glasses, promised for the coming months.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made with our newsroom’s technology tools, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is not published automatically; it is held for the editor, who decides whether it is fixed, published or dropped.
- Archive story
- It was written after the event. Its dates were checked against the date of the event, including the dates of any later developments it reports.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
Companies in this story
