Automation credit usage
How Automation runs use AI credits, and how to check and reduce usage
Automation runs use the same AI credits as the chat. They come from your team's shared pool, and there is no separate charge for Automation. For how AI credits work, see AI credits.
How usage is calculated
Usage depends on the kind of step that runs.
| Step | Usage |
|---|---|
| Pipeline (bringing in data) | 1 credit per 10 minutes of runtime |
| Query, Agent, Notify (steps where the AI works) | Like the chat, based on how much the AI does and which model it uses |
- Pipelines are proportional to actual runtime. There is no rounding up: 3 minutes is 0.3 credits
- Steps where the AI works use more when the data or instructions are longer
- Creating or changing an Automation in the chat counts as regular chat usage
Examples (pipelines)
| Schedule | Runtime per run | Per run | Per month (30 days) |
|---|---|---|---|
| Once a day | 3 min | 0.3 credits | 9 credits |
| Once a day | 25 min | 2.5 credits | 75 credits |
| Every hour | 2 min | 0.2 credits | 144 credits |
If the Automation also has query, agent, or notify steps, their AI usage is added on top.
Checking usage
- Per run — select a run under Recent runs on the Automation screen to see its Credits used
- This billing period — the Automation card under Settings > AI Usage in the sidebar shows usage, pipeline runtime, and the number of runs for the current billing cycle
Automation and the chat share the same pool. Many frequently scheduled Automations leave less for your team's chat. If you run low, you can buy more credits.
Keeping usage down
- Load data one day at a time — only new data is fetched instead of everything, so runs are shorter
- Review how often it runs — running every hour uses 24 times as much as once a day. Match the schedule to how often the data changes
- Be clear about when to send — "only send when the condition is met" means the AI writes messages less often