azure Microsoft Azure Blog ·

Azure Enhances AI Agent Cost Governance and Optimization with Foundry and API Management

blogazurepreviewarchitectfinance
feature announcement

Microsoft details strategies and capabilities for optimizing AI agent costs and governing their spend within enterprise environments. The focus is on providing visibility, setting limits, and measuring ROI for agentic systems built on Microsoft Foundry and Azure. Key features include cost attribution in Foundry, token usage monitoring, project-level cost tagging (in preview for Azure OpenAI), and API Gateway-enforced token rate limits and quotas. These controls help IT leaders, FinOps teams, and developers manage consumption and prevent unexpected cost increases as AI agents scale.

  • Enhanced Cost Visibility for AI Agents in Foundry and Azure
  • Enforce Token Rate Limits and Quotas in Foundry
  • Apply LLM Token Limit Policy Across Diverse Model Providers
  • Integrate Financial Budgets with Microsoft Cost Management
  • Measure Agent Value and ROI Through Observability
Features (3)
  • Enhanced Cost Visibility for AI Agents in Foundry and Azure

    Microsoft Foundry now provides capabilities to see estimated costs, inspect token usage for individual agents, and monitor model costs, including project-level cost attribution via tags. Azure API Management's AI Gateway also emits token metrics, while Foundry tracing captures tool usage, retries, latency, and token consumption to explain agent behavior.

  • Enforce Token Rate Limits and Quotas in Foundry

    Foundry Control Plane, when configured with AI Gateway, can enforce tokens-per-minute rate limits and total token quotas at the project scope. These controls happen in the request path, rejecting requests with 429 or 403 responses.

  • Apply LLM Token Limit Policy Across Diverse Model Providers

    The `llm-token-limit` policy can limit token consumption per key (e.g., subscription, team) using rate or cumulative quotas. It applies consistently across OpenAI-compatible APIs, Anthropic Messages API, and other backend servers.

Enhancements (1)
  • Integrate Financial Budgets with Microsoft Cost Management

    Microsoft Cost Management budgets provide an authoritative view of actual and forecasted spend, using Azure billing data and real prices. While not instant caps, they serve as accountability and escalation tools, connecting to Azure Monitor action groups for custom workflows.

Notes (1)
  • Measure Agent Value and ROI Through Observability

    Understanding agent value and return on investment involves continuous evaluation of quality, safety, groundedness, and task completion. Observability signals help teams determine if rising costs are due to demand, inefficiency, or regressions, enabling informed decisions on model choice and early issue detection.

Read the original announcement →

https://azure.microsoft.com/en-us/blog/the-economics-of-agent-optimization-how-ai-agent-governance-controls-cost-and-proves-roi/

Related releases