How Postman runs Agent Mode for 40 million developers on Amazon Bedrock
TL;DR
Building an AI agent that works in a demo is a different problem from running one for 40 million developers. Postman and AWS share the architectural patterns behind Agent Mode: controlling tool sprawl, exposing schema-based reads, and treating context as the real bottleneck, plus how it runs on Amazon Bedrock at scale.
Nauti's Take
Teams building a similar agent should start with a focused load test: How many tools does a typical task require, how quickly does context grow, and what does a failed run cost? The AWS post offers useful architectural guidance, while independent figures for latency, cost, and failure rates remain unavailable.
A small team should measure those variables before rolling the pattern out widely.
Sources
