Meta AI Debuts Mac App With Screen Reading and Business Tools

Meta is bringing its AI assistant to the Mac with a new desktop app that reads whatever is on screen and takes dictation across the whole system, while a parallel update wires the assistant into the company's advertising and productivity stack for business users.

What happened

The new macOS app gives Meta AI system-wide dictation, letting it type into any application - a move that puts it directly against standalone dictation tools such as Wispr Flow, Superwhisper, and Monologue. It can also look at the current screen and answer questions about what is visible, powered by Meta's Muse Spark model. Google made a comparable move last month when it enabled system-wide dictation in its Gemini app for Mac, according to TechCrunch.

The launch ships alongside a Meta AI update aimed at business owners. Merchants can now connect their Instagram and Facebook accounts, Meta ad campaigns, and Google Workspace accounts - Gmail, Docs, Sheets, and Slides - to Meta AI and query the assistant directly.

Why it matters

Once connected, Meta AI can surface campaign performance and audience engagement metrics, explain which posts performed well, and pull competitive intelligence from publicly available data. It can also draft proposal decks, documents, and spreadsheets from those connected tools, giving small businesses a low-friction way to automate parts of their marketing and back-office work.

The integrations put Meta's stated ambitions into practice. On the company's second-quarter earnings call, CEO Mark Zuckerberg called selling agents to businesses and automating work on their behalf a "big opportunity" - and the new connections turn Meta AI from a chat assistant into a productivity layer that spans consumer and business use cases.

For AI builders, the update is another signal that major model providers are converging on the desktop as the next battleground. Meta, Google, and OpenAI are all racing to make their assistants context-aware and capable of acting across a user's apps - the same agentic shift playing out in agentic software development. Each is positioning itself as the default AI layer for individuals and the businesses that depend on them.