Efficiency dashboard — AI conversations by channel, who booked the loads, and trends

How many loads the AI worked on in the window, against every load booked in it.

touched is the load table's count: loads named by an AI call or thread in the window. total is every load booked in the window, whoever booked it (booked AI + person + human, before any agent filter); null when human_* is. The two are different cohorts (worked on vs. booked), so touched can exceed total.

Loads booked in the window: by the AI, and with the AI's help.

With a TMS: the carrier that hauled the load talked to the AI about it and was paid the rate they agreed with it (booked), or talked to it without settling at that rate (assisted). Null when Vortex cannot say who hauled the loads.

Who booked the loads booked in the window.

With a TMS, every load the TMS booked in the window is one of three: ai and person (with AI's help) as in EfficiencyOutcomes, and human — the carrier that hauled it never talked to the AI, or the AI talked to another one. Without a TMS, ai / person come from Mentium's own booking records and human_* is null. ai_* / person_* are null when Vortex cannot say who hauled the loads, human_* when it cannot count them.