What is the Critic?

If you’ve seen our posts mention “the critic said no,” this is who they’re talking about. The Critic is the quality gate of the whole system: an agent whose only job is to judge the work of the other agents — and reject it.

What does it check?

Every post gets an overall score from 1 to 10, built from six weighted categories:

• Brand voice (20%) — does it sound like you, or like generic AI copy?
• Hook strength (20%) — does the first line earn a stop mid-scroll?
• Call to action (15%) — is there a clear next step for the reader?
• Grammar (15%) — correctness in your language, including accents and diacritics.
• Hashtag quality (15%) — relevant and platform-appropriate, not spam walls.
• On-screen text (15%) — the text that appears on the image itself.

Anything below your threshold is rejected. A rejected post doesn’t quietly disappear — the Critic’s feedback goes back to the writer, which revises and resubmits. Still failing? The post lands in your review queue instead of your feed.

Does anything look at the actual image?

Yes. A separate visual reviewer inspects the final rendered image — the actual pixels, not the plan. It checks that on-image text is readable and spelled exactly right, that there are no AI-generation artifacts, and that the subject matches what the caption promises. If it finds a problem, the image is regenerated with corrections; if it’s still not right, the post is held for your review rather than published.

Why a critic at all?

Because generation is easy and judgment is rare. Any tool can produce ten captions in a second. The hard part is having something that says “this one isn’t good enough” before your customers see it. Every score is also stored — category by category — which is how the Learner improves the system week over week.