Spend

The Spend tab on Home estimates what your organization spends on AI and shows the token usage behind it. It attributes spend to seat-based subscriptions, metered token usage, and usage beyond a plan's included allowance, so finance and platform teams can see where the AI budget goes.

Note: Spend is a Beta tab and is enabled per tenant. If you don't see it, contact your WitnessAI representative.

The Spend dashboard on the Estimated spend tab: Estimated spend, Subscriptions and Metered usage cards on the left, and the Spend overview with token cards, Spend over time, Tokens by model and Tokens by user on the right.

How it works

WitnessAI counts the tokens that flow through each AI application and model it observes, prices them, and combines them with the subscription plans you describe in Spend settings. Seat-based plans contribute a fixed cost per seat. Usage within a plan's included allowance is covered by that cost. Usage beyond it, along with traffic from applications that aren't on a plan, is priced at metered rates.

Spend took over token cost reporting from the Total Cost view that the Value tab used to have. Value now focuses on risk avoided, cost saved and AI impact.

Every figure is an estimate. For a model that has no recorded cost estimate, WitnessAI uses the default rates you set. While those rates are $0, that usage is counted in tokens but carries no dollar figure.

Open the dashboard

Select the Spend tab on Home. The dashboard opens on the Estimated spend tab. The header holds:

Headline: a sentence such as "Your AI stack cost $X this period: N% subscriptions, N% metered." When there's usage beyond plan allowances, the split also shows the share of extra usage.
Date range: choose Last 30 days (the default), Last 60 days, or Last 90 days.
Settings gear: opens Spend settings.
Updated timestamp: when the data was last refreshed.

The tabs

The cards on the left switch between three tabs, each showing its total for the period.

Tab	What it shows
Estimated spend	The Spend overview: total, input and output token counts; Spend composition (how spend splits across subscriptions, extra usage and metered usage, shown once you have subscription plans configured); Spend over time; Tokens by model; and Tokens by user.
Subscriptions	Spend on seat-based plans. With no plans configured, the tab shows No subscription plans configured and a link to Spend settings.
Metered usage	Cost or tokens by model, by application, and as an application trend. The model chart counts all traffic at API-equivalent cost, including plan-covered usage. The application charts leave out applications covered by a subscription plan.

The Tokens by model, Tokens by user and application charts have a Top N selector, so you can widen or narrow the list.

Export chart data

Every chart on the Estimated spend and Metered usage tabs has an Export CSV control in its header. Tokens by user exports up to the top 10,000 users by token volume for the selected range. The other charts export their full data series.

Spend settings

Select the settings gear to open Spend settings. It has two tabs.

Tab	What you set
Subscriptions	The seat-based plans your organization pays for. Turn on a preset plan (ChatGPT Business, ChatGPT Enterprise, Claude Team, Claude Enterprise, Cursor Teams or Microsoft Copilot) and add its seat types, with the number of Seats, the $ / seat / mo price and the Included M tok / mo allowance where the plan has one. Use Add custom subscription for a plan that isn't listed.
Metered rates	Default rates ($/1M tokens) for Input and Output tokens, used when a model has no recorded cost estimate.

Select Save configuration to apply your changes, or Reset to defaults to set the default rates back to $0. View methodology & sources explains how each figure is calculated.

Note: Only the Admin role and the roles above it can save Spend settings. Other roles can open the drawer to read the values, and see "Contact your administrator to update the spend configuration."

Limits

Spend is a Beta feature and is enabled per tenant.
The date range is limited to the last 30, 60 or 90 days.
Model rows on the Metered usage tab can't be split by application yet.

Spend How it works Open the dashboard The tabs Export chart data Spend settings Limits