API
< Back to all clusters
[TECHNOLOGY] · France · 2 sources

started · updated

AI expense governance: managing autonomous agent costs

Companies are facing increasing challenges in managing the costs associated with artificial intelligence, particularly regarding autonomous AI agents and decentralized software usage. AI agents that operate on a per-token or per-call basis can rapidly exceed budgets if instructions change or processes become inefficient, often impacting the margins of software providers who sell agents at fixed rates.

An IDC survey indicates that 67% of surveyed companies exceeded their computing and model budgets for AI agents by more than 10% over a twelve-month period. This issue is compounded by "shadow AI," where employees use personal accounts or corporate cards for individual subscriptions without centralized oversight. An Ipsos bva study noted that while 35% of employees use AI weekly at work, only 21% have received formal training.

To mitigate these risks, experts recommend implementing strict expense governance. This includes creating a centralized inventory of AI licenses, establishing authorization protocols for paid subscriptions, and setting usage caps. Effective management requires tracking tools by team and purpose to identify redundancies and ensuring that sensitive data is only processed through approved, company-sanctioned tools.

Entities

Google · IDC · Ipsos bva