Keep the weekly review running before Notion's AI reset

The weekly review is where a metered AI tool shows up: the notes are late, the agent is paused, and someone asks why the dashboard says the pool is empty. The fix is a short operating check that turns the credit pool into a line item before the month starts.

Notion began applying the new AI credit system to newly created Business and Enterprise workspaces on October 19, 2026, and applied it to existing workspaces on November 2, 2026. For a small team, that date is the deadline for the weekly review, not the start of the month.

The new system allocates 1,500 Standard credits and 500 Select credits to each workspace member every month. Notion treats Standard and Select as distinct per-member balances; they reset each month, unused amounts expire, and members cannot merge or move them to another person. Heavy use by one user does not create a shared surplus, so the team needs a split before the first meeting consumes the pool.

The meter changes the review, not the meeting

The meeting still needs the same outputs: decisions, owners, open risks, and next actions. What changes is the cost of producing them. Auto-routed simple, medium, and complex tasks are estimated at roughly 5, 10, and 25 credits. Several complex prompts can burn a meaningful share of the pool before anyone talks about the roadmap. Use AI, but know which ritual owns the spend.

AI meeting notes will start drawing on Standard credits on December 1, 2026, and a one-hour meeting is estimated at roughly 10 Standard credits. A weekly review that uses notes makes the feature a recurring line item, not an occasional convenience.

The pre-reset check is the control point

Run the check before the reset, not after a pause. Each item is testable in a minute.

• Confirm the plan and reset date. The usage dashboard shows how many credits have been consumed and when the monthly pool refreshes for the selected period. A missing date means the review is not ready.
• Split the pool by ritual. Assign Standard and Select usage to the meetings, docs, and agents that actually consume credits. The split shows which rituals draw on Standard and which draw on Select.
• Name an owner for each agent. The dashboard separates runs and credit consumption by agent, so operators can see which agent drives spend. The owner needs to say what the agent does, who approves changes, and what happens when it pauses.
• Set a buffer decision. Notion credits purchased for a workspace cost $10 per 1,000 credits per month. Notion sends in-app and email alerts to admins when a workspace reaches 80% and 100% of its credit usage. Set the buy trigger before the 80% alert.
• Run the short usage check. The usage table has a member-spend row that identifies which users exceed the AI allowance and how much they spend on high-credit models. Flag any out-of-line name for the owner before the next review.

Custom agents add a pool split. Agents created before November 2, 2026 continue to draw on workspace Notion credits, while agents created on or after November 2, 2026 draw on Select credits. The same team can have agents doing similar work while they hit different pools. When a workspace lacks enough credits, Notion pauses Custom Agents that rely on shared workspace credits until the pool resets or an admin adds credits. The agent owner tracks which pool their agent uses before a pause.

The fallback should be boring

The fallback is the manual version of the ritual: a template, a short prompt, a human summary, and a place to paste the output. An agent pause does not stop the meeting from producing the same document. Hand-written action items cover unavailable notes. The meter changes the path, not the output.

Keep the checklist in the same place as the weekly review. The habit is small: open the dashboard, note the owner, and confirm the fallback before the meeting starts. A healthy pool lets the review run; a thin pool sends the owner a note first.