ServiceNow meters AI two ways: flat account-level Assists and capacity-based AI systems

Role-based counting can be audited from a table the customer controls. The platform counts active users who have logged in within the past 365 days, are not flagged as Web Service Access Only, and hold a Fulfiller role mapped to an entitled application ^[6]. Take the role away and the next measurement drops the user. Now Assist's Assists entitlement works the other way: the guide describes that category as needing no user counting and no discovery queries, with the entitlement active once bought ^[2].

The guide was posted on ServiceNow's community site by an author who says they worked through every ServiceNow licensing reference page and knowledge article available before compiling a workbook ^[19]. On the Assists model it is direct. Consumption "can spike faster than contracts anticipate, and unlike user-based models there's no natural ceiling built into headcount," the guide said ^[4].

AI turns up in two of the three categories. Assists sits at account level ^[3], and "AI systems" appear in the list of discovered units that capacity-based products measure, alongside servers, devices, managed software, nodes, robots, portal visits and work orders ^[13]. Two of the three meters can therefore bill for AI work, and neither is bounded by the number of people on the payroll ^[20].

Role assignment matters for the same reason. A user qualifies as a Business Stakeholder if they hold an Approver role and no Fulfiller role, and the measurement crosses two tables, one for role assignment and one for approval activity ^[9]. The moment that user gains a Fulfiller-mapped role, they leave BSH measurement entirely ^[10]. Products such as HRSD and Employee Center Pro count every active user on the instance regardless of role ^[11]. On that model, the guide said, "tenant-wide active user governance stops being a hygiene task and becomes a direct cost control" ^[12].

Capacity products bill the peak count in the most recent accrual period, and a single-day spike from a large discovery run can set the number for that period ^[15]. The count can be checked in-instance: open the usageanalytics_count table, add the DefinitionID field through the column chooser, sort by most recent date descending and filter for the relevant Definition ID; if a definition appears inactive, check usageanalytics_count_cfg and then ua_stats_defn ^[16]. ITOM alone has more than twenty active definitions covering different SKU combinations ^[17]. Knowing which Definition IDs a contract actually covers is, the guide said, "non-negotiable before any renewal conversation" ^[18].

The difference between the meters is what a correction can undo. Removing a role changes the next role-based count ^[6]. A discovery run that has already set the peak for an accrual period fixes the billable figure for that period ^[15].

The decision in front of most ServiceNow customers this quarter is narrow: reconcile Definition IDs and role tables against the contract before the renewal meeting. The longer question is whether Assists grows into the largest item on the bill. The guide does not include rates for any of the three categories, or define how long an accrual period runs ^[22].