Category Allocation
How monthly AI spend is split across Hannibal, Marcus, Constantine, Cato, and H1 Triage, and what a 0% slice does to automation.
Category Allocation
Category Allocation divides this tenant's monthly AI cap across five use-case buckets. It does not change the cap itself.
The five categories
Each enabled category shows a usage tile. Categories with no allocation appear disabled/gray.
If the tenant has never been provisioned for a customer budget, the card is display-only: Not provisioned yet. There is no Edit.
Who can edit
Anyone with manage settings (a customer Admin). That is deliberately looser than Monthly AI Spend, which is super-admin only. Admins split the pie; only super-admins change the pie's size.
Percentages are whole numbers and must total 100%. Presets and arrow-key nudges keep the total fixed: a 1-point increase in one category is offset by redistributing that 1 point across the others. Saving writes every category in one request; the backend requires an exact sum.
Turning allocation off
Disabling allocation deletes every category limit and returns inference to the shared platform budget. That is not the same as setting a category to 0%.
What 0% does to automation
A 0% Hannibal slice while Automatic Penetration Tests is on does not block saving allocation. The UI shows that scheduled hunts will pause. The hunt card then shows a paused state because Hannibal's max_limit is $0. The hunt switch itself is not flipped off.
A 0% Cato slice while Automatic Findings Validation is on behaves the same: Cato pauses at $0 headroom; the validation switch stays as you left it.
What this card never writes
It reads the hunt-schedule and Cato settings only to word that pause notice. It never enables or disables those features.