AI usage limits

Three caps, not one

Every plan includes a monthly allowance of AI work, split into three buckets:

• AI drafting — making new post copy out of something that was not post copy yet: an idea, a prompt, a photo, a clipped article.
• AI rewrites — reworking copy that already exists: adapting a post for another platform, polishing, generating variations.
• Post idea generation — producing ideas rather than post copy.

They are separate on purpose. Running out of idea generation does not stop you drafting, and heavy rewriting does not eat into your drafts.

Counted per action, not per model call

One "draft this" is charged as one draft, even though it runs several passes underneath — voice matching, a platform shaping pass, a critique pass. Charging three would make the published number meaningless.

The counters are per account, not per user and not per workspace. An agency on one bill has one pool. Everyone on the team draws from it, and no client workspace can be starved by another's usage because there is only the one pool.

What is not counted

Work you did not ask for is never charged to you:

• The weekly post-ideas email digest
• Learning your voice from your past posts and your edits
• Working out what a message you sent means before acting on it
• Content moderation

Metering those would let a background job spend an allowance you never chose to spend.

Which buttons spend which bucket

Generated images are separate

AI image generation has its own monthly allowance, counted in images rather than in actions, and it is not drawn from any of the three buckets above. See AI image studio for the per-plan credit numbers and what happens when they run out.

When you hit a cap

The action is refused before any work is done, so a refusal never costs you an action out of next month's allowance. You get a message naming which cap you hit and what your options are — the next tier up, or waiting.

Nothing else stops. Scheduling, publishing, analytics, approvals, the inbox, and anything you write yourself all keep working. Only the AI generation paths pause.

At 80% of a cap DemandBird starts warning you, so running out is not a surprise.

When it resets

At the start of each calendar month, in UTC. Unused allowance does not roll over.

Raising a cap

Upgrading takes effect immediately — the higher cap applies to the rest of the current month, it does not wait for the next one. Usage already counted stays counted.

If your usage pattern does not fit the published tiers, talk to us. We can set a custom allowance on an account without moving you to a plan you do not otherwise need.

Where the enforced numbers live

The caps above are read straight from the same source as the pricing page, and the app enforces them on the server at every entry point — the editor, the API, MCP, and SMS all pass through the same counter. There is no path that spends an allowance without recording it, and no path that records without the cap applying.