Meta for Business’ Post

Seeing Muse on the integration list means your favorite coworker is ready to launch. 🤝

What's the difference between Spatial Ambient AI and Muse AI?

Interesting direction for small businesses. The approval layer is essential, particularly when an agent can connect content, operations and ad accounts. The value will depend on whether Muse can preserve commercial context across tools—margin, stock, lead quality and brand rules—rather than optimising isolated tasks. Automation should reduce operational friction without removing human accountability.

I just started using Muse personally, and what an achievement it is. The best agent I've seen in action.

The most important design choice here is 'you're always the approver.' For agents that can publish, send, or spend, human approval on consequential actions is what makes small businesses comfortable delegating at all. Connecting to the tools teams already use, rather than asking them to move, also lowers the barrier a lot. The real test will be how well an agent learns a brand voice and priorities over time, and how easy it is to review what it did. Interesting launch for anyone running a lean team

The approval boundary is the part I’d watch most closely. For a small business, the useful question isn’t how many applications an agent connects to. It’s whether it can remove one complete recurring workflow without creating another checking job. Pick something measurable, lead follow-up, invoice chasing, weekly reporting, establish the effort today, then see whether the whole workflow actually takes less time.

"You ruined a system that was actually working fine. While boasting about AI and new platforms, you are ruthlessly closing people's and small businesses' accounts without any proper support or reason. Fix your broken moderation and account support system first before pushing new services!"

What interests me most is the distribution logic: instead of asking SMBs to adopt another platform, Muse plugs into the martech stack they already pay for — Klaviyo, Shopify, HighLevel. The agent that wins won't be the smartest model, it'll be the one embedded where the work already happens. Curious how attribution works when the agent touches every step of the funnel.

The move from AI assistants to AI agents that can actually work across existing business tools is a pretty significant one. Keeping the human as the approver makes that model even more practical.

I assume Amazon is smart enough to identify this as a web-scrolling/automation tool and blocked it as per their policy. Meta for Business, it’s time to push your Product Management team to think outside the box, come out of the video-game world, and start designing products that solve real-world problems! If you are paying your product team around $50,000 per annum, then Muse may be worth the deliverables. But if you are paying them more than $200,000 a year to design products like this, then you seriously need to question the leadership, product strategy, and whether shareholders’ money is being used effectively.