AI calls cost money at your provider. KanbanWP gives you three controls: who may use AI, how many tokens each person may use per month, and a log of every call.
Choose who can use AI
On KanbanWP > Settings > AI, the Allowed roles list shows every WordPress role on your site. Tick the roles that may run AI features.
- A user can use AI if any of their roles is ticked.
- KanbanWP administrators can always use AI, whatever you tick.
- A user who is not allowed gets a “You are not allowed to use AI features” error.
- AI also needs a configured provider and key. Without them the request fails with “AI is not configured.”
Out of the box, the Administrator and Editor roles are ticked.
Running an AI feature on a board also requires edit access to that board, so a user who is only a viewer on a board cannot use AI on it. See permissions and roles.
Set a monthly token cap
In the same tab, Monthly token cap sets the total input plus output tokens each user may spend per month.
- The cap is per user, not shared by the whole site.
0means unlimited.- The default is 1,000,000 tokens per user.
- Usage counts from the first day of the calendar month, using UTC.
Before each request, KanbanWP adds up the user’s tokens this month. If they have reached the cap, the request is refused with “Your monthly AI token budget has been reached.” The check happens before the call, so the call that crosses the line still completes, and the next one is refused.
The cap resets on its own at the start of the next month. To give someone more room sooner, raise the cap, which applies to everyone.
Read the usage log
Go to KanbanWP > AI Usage. You need the KanbanWP manage permission.
The screen has two tables:
This month, per user lists each user with Tokens used and a Cap usage bar with a percentage. The bar turns red at 100 percent. With no cap, the bar is replaced by a dash.
Recent AI calls shows the latest 100 calls with Date, User, Board, Feature, Provider, Model, Tokens in and Tokens out.
The page also reminds you of the current cap and where to change it.
Practical advice
- Start with a cap that matches your provider budget divided by the number of AI users.
- Check the screen after the first week to see which features use the most tokens.
- Set a spending limit with the provider as a second safety net, since the token cap counts tokens, not money.
- Use a cheaper model in the Model field for routine work.
The log covers AI calls made through KanbanWP’s AI features. It does not include external agents that use the MCP server, because those bring their own model and billing.