Workspace and billing
Usage and answer budgets
One answer is one engine’s response to one prompt. Everything Ocita measures is counted in answers.
What a brand uses
A weekly run uses active prompts × engines × repeats. With 50 prompts, the most a brand can have, on four engines at 5 repeats, that’s 1,000 answers a week, or about 4,350 a month. A brand’s first check uses 3 repeats.
Prompts shows what the current set uses each week, and warns when it would go over the brand’s monthly allowance. Pausing prompts, turning off an engine, or moving to 3 repeats in Brand settings uses fewer.
Included answers
Each brand includes 4,350 answers a month on Standard, enough for 50 prompts at 5 repeats, and 2,175 on Founding. Months run from the 1st, in UTC. Unused answers don’t carry over.
Usage and billing shows:
- A forecast of how much of your included answers you’ll use this month at the current schedule.
- Answers by brand, with Near cap beside any brand above 95%.
- Your credit balance.
Portfolio shows each brand’s usage too.
A brand’s monthly cap
In Brand settings, Answer budget sets the brand’s Monthly cap and what happens At the cap. Only Owners and Admins can change them.
The plan’s included answers come first. The cap can be set lower, to use less, or higher: answers above the included amount, up to the cap, come from the workspace’s credits. If credits run out, the run pauses.
- Keep running on credits and notify me: the run goes ahead, and answers past the cap come from credits too.
- Pause runs until next month: runs that would go past the cap don’t start.
Either way, people who can see the brand get an email if they’ve turned on Budget warnings, and Activity records it when a run starts using credits.
The cap counts every answer from weekly runs and the first check, whichever paid for it. Check now always uses credits and doesn’t count against the cap.
Spending limits on our side
Ocita has a daily limit on what engine calls can cost across all workspaces, and a switch to pause all runs. If either is on, new weekly runs wait and start once it clears, within the same week, and Check now is refused without taking credits.