Every Productive plan includes a monthly amount of AI usage, shared by everyone in your organization, including agents. When your team uses Productive AI, for example chatting with the AI Assistant or running agents, it draws from this shared amount.
đ Need more? See Increase Your AI Usage Limit.
What's Included and When It Resets
It depends on your plan and team size. How much is included depends on your subscription plan and the number of seats you pay for.
It resets on the 1st of every month (at midnight UTC), whatever your billing date is. Whatever your team doesnât use by then expires and doesnât roll over to the next month.
You can see how much your organization has used in Settings > Usage. See Tracking Your AI Usage below for more details.
What Counts Toward AI Usage
These count toward your organizationâs AI usage:
Conversations with the AI assistant, including assistant triggers and the artifacts it creates
AI agents, whether you chat with them or they run on a trigger
Meeting notes and action points for Notetaker recordings
đ Most of your organizationâs AI usage comes from the AI assistant and agents. The other features count for very little.
These donât count toward your AI usage:
MCP Server, because the AI model runs in the tool you connect, such as Claude or ChatGPT
Dictate message, because only the message you send to the AI Assistant counts
If an AI run fails or you stop it, it doesnât count toward your usage.
â
đ Notetaker recordings and transcripts donât count toward your AI usage. They count toward your organizationâs monthly Notetaker hours. Read more in Buy More Notetaker Hours.
What Makes a Request Use More
Not every request uses the same amount. The more work the AI does, the more it uses. These things make the biggest difference:
The model. More advanced models use more for the same request. The default model handles most everyday work well.
Thinking effort. Higher thinking effort means the AI reasons longer before it answers, which uses more.
How much it has to read. "Whatâs due today?" uses less than "Summarize all 40 active projects."
How many steps it takes. Looking up one task uses less than finding 20 tasks, comparing them, and updating each one.
How often it runs. Agents and assistant triggers run on their own, so a small task that runs every hour adds up much faster than the same task run once a week.
đ Admins can choose which models everyone in your organization can use in Settings > AI. Learn more in AI Models.
Tracking Your AI Usage
Go to Settings > Usage and find the AI usage card. It shows how much of this monthâs AI usage your organization has used as a percentage.
The Usage page shows your organizationâs total, not a breakdown by person or feature.
đ Only users with billing permissions can see the Usage page.
When Your Organization Reaches Its Limit
At 80%, a banner appears in the AI Assistant chat to let everyone know the organization is getting close. Admins also get an email.
At 100%, AI features are paused for everyone in your organization until usage resets. Admins get a second email. Agent and assistant triggers still fire, but their runs fail and the run log shows AI usage limit reached. Once usage resets or you add more, new runs work as usual. Runs that failed at the limit donât run again.
Billing admins see an Upgrade now button in the banner that takes them straight to the AI usage package.
Everyone else sees who to ask: the banner names your organization owner.
đ To add more, see Increase Your AI Usage Limit.
đ Meeting notes and action points are still added to new Notetaker recordings, so your usage can go slightly over 100%.