A small-business tester had Muse compare recent Facebook Page post performance, pick a format, write a caption in her voice and, in a follow-up test, publish to her Page and to a Group she administers. She says the analytics it reported matched what she saw in Meta's own tools.
- First reported
- aiagentslibrary.com · 10 Oct 2026
- Added
- 11 Oct 2026
- Source checked
- 11 Oct 2026
What they did
The founder of the review site AI Agents Library tested Muse on her real Facebook Page and a public Facebook Group she administers. She first asked what Muse could see and do on the Page without acting, then for her best and worst recent posts using only figures from its Facebook connection. She then had it write a caption modeled on one of her earlier posts, make an image, schedule the post, and post in the Group after she signed in to Facebook through Muse's browser.
What happened
She says the analytics matched what she checked in Meta Business Suite, the caption sounded like her, and both posts went live. Images improved sharply once she uploaded one of her own designs as a reference; she rated the whole workflow 8.2/10.
Keep in mind
A typo inside the image reached the live Page post, Muse could not see her Page images behind Facebook's login, and at one point it said her Group did not exist. Group posting needed her own sign-in, and she checked each step.
Try it with your agent
Look at the posts on <your business page> from the last <number> days and rank them by views, link clicks and comments, using only the numbers you can retrieve; write "not available" for anything you cannot. Tell me which format worked best and which one my audience has not seen recently. Draft a post about <topic> in the style of <link to one of my past posts>, suggest an image, and schedule it only after I approve the final text and image.
Replace the <angle brackets> with your own details.
Original source
aiagentslibrary.comHands-on reviewA journalist or reviewer used the agent and wrote up what happened.
