## **📊 Why credits disappear in the first days**

Atlas agent playbooks draw from your AI Agent credits based on the AI usage each run generates. An hourly routine can start many times in a day, so the same playbook can generate usage charges over and over. If several playbooks are set to hourly, usage adds up quickly and the monthly balance can be gone after the first day or two. The agent then has nothing left to spend until credits renew, which is why it can look stopped for the rest of the month.

Credits are shared across Atlas agent activity. They are not reserved separately for later in the month. Early, frequent runs simply use the pool first.

## **🔍 What to check before you change anything**

Review the agent schedule and recent activity so you can see which routines caused the spike:

- **Frequency:** Look for playbooks set to hourly rather than daily or weekly.
- **How many playbooks are active:** Each active routine spends credits on its own schedule.
- **Run history:** Note how many times each playbook ran in the first days of the month.
- **Credit balance:** Confirm whether the balance hit zero and when it is set to renew.

A useful estimate is runs per day times days active times the credits used per run. Hourly scheduling is the setting most likely to exhaust a monthly allowance early.

## **🛠️ How to make credits last the month**

1. Pause any hourly playbook you do not need running continuously.
2. Change remaining routines from hourly to daily or weekly, based on how fresh the output needs to be.
3. Keep only the playbooks that support current work active. Turn off duplicates or tests.
4. Check the credit balance after the next scheduled run so you can see the new pace.
5. If the balance is already at zero, wait for the monthly renewal or ask support to review whether a plan or credit adjustment applies to your account. Changing the schedule now prevents the same spike after renewal.

Hourly routines are best for short tests. For ongoing SEO work, a slower schedule usually gives you coverage all month without a large early charge.

## **💡 Reviewing September and October AI charges**

Usage is tied to playbook runs, not to a single end-of-month action. September and October charges can look high if hourly routines were left on at the start of each month. Gather the playbook names, the schedule each one used, and the dates the balance dropped. That context lets support match AI charges to specific agent activity.

Support can walk through those charges with you. They cannot be reconstructed from this article alone, because balances and plan details differ by account.

## **🚀 Prevent the same escalation next month**

- After any new playbook is turned on, watch credit use for the first 24 hours.
- Avoid leaving test routines on an hourly schedule.
- Set a reminder a few days after renewal to confirm the balance is declining at the pace you expect.
- If the agent stops working, check the credit balance before assuming the agent itself failed.

If you need further assistance, open the chat widget in the bottom-right corner of the platform and send your message. A member of our support team will reply in the chat.