Control AI Agent Costs with Token Limits | Postman Fabric Gateway

How do you control token usage when multiple teams are building and running AI agents across your organization? In this video, I demonstrate how to enforce token limits centrally using Postman Fabric Gateway. You can apply these limits to an LLM service or a specific route, with conditions that determine which requests the policy covers. Using a banking support agent, I set a budget of 10,000 tokens per 60 seconds and compare what happens when the agent connects directly to its dependencies versus routing requests through Fabric Gateway. Once the gateway-routed agent exhausts its token budget, requests return a 429 response. You'll learn how to: • Configure the token limit plugin. • Apply token budgets at the service or route level. • Match policies to specific paths and HTTP methods. • Identify what happens when an agent exceeds its token budget. #Postman #FabricGateway #AIAgents #LLM