Microsoft has launched Agent 365, a platform designed to detect, manage and protect AI agents used inside companies, including agents running in Copilot, Teams and Microsoft 365 as well as partner agents. The company says the spread of AI agents is creating security risks as employees deploy tools outside IT oversight. Agent 365 adds detection via Defender and Intune, multi-cloud discovery through AWS Bedrock and Google Cloud, and new mapping and lifecycle features.

As AI agents spread, usage-based pricing and performance-based models that charge only when AI completes work properly are rapidly gaining ground in the enterprise software market. The industry has long relied on flat-rate subscriptions priced per user per month, but companies increasingly see those plans as a poor fit for AI. Some firms are also moving beyond volume-based models, adopting outcome-linked pricing. Adobe plans to apply performance-based fees to its CX Enterprise AI agent platform, while keeping usage and subscription pricing for some existing AI tools. Zendesk, Intercom, Sierra AI, Salesforce and HubSpot have also introduced outcome-based approaches.