AI Admin Connectors

Generative AI adoption in most organizations happened from the bottom up. Teams and individual developers signed up for their own models and subscriptions, and IT ended up without a list of what was running or what data was going into it. Finance got the other half of the problem: consumption bills that moved every month, with no way to see which team generated them.

The new connectors read data directly from the AI provider admin APIs. What comes across is token usage broken out by user and by model, with the cost attached, landing on a normalized schema so the numbers are comparable across Anthropic, OpenAI, Cursor, and Snowflake instead of sitting in four separate admin pages with four different shapes.

Once that data is in Domo it behaves like any other dataset. The user-level detail means spend already has a name attached, and joining it to your user directory in Magic ETL rolls it up to a team, a department, and a cost center. Visualize month-to-date spend by department, set an alert when a team crosses its budget threshold, or break the same data down by model to find the teams running an expensive model for work a smaller one would handle. Pull it into a data model alongside the rest of your SaaS spend and AI stops being the one line on the invoice nobody can account for.