Postman details how Agent Mode serves 40 million developers on Amazon Bedrock
AWS outlines the architecture behind Postman's Agent Mode, which the company says serves 40 million developers and runs on Amazon Bedrock. The account describes narrowing the agent's toolset, shaping application context, and requiring approval for actions that change application state.

Full Report
AWS published its account of Postman Agent Mode's architecture on October 9, 2026. Postman says Agent Mode serves 40 million developers and runs on Amazon Bedrock. The company says managed access to foundation models through Bedrock lets it scale without operating its own model-serving infrastructure.
Postman identifies tool selection and context preparation as central design challenges. In its testing, tool-selection errors increased when the visible toolset exceeded approximately 40 tools. Its root agent therefore narrows a catalog of more than 170 tools to about 15 relevant tools for each request. For structured data, the system uses schema-based queries; dedicated context handlers distill application entities, while a retrieval-augmented knowledge base selects feature articles based on the query and available context.
The reported controls include user approval before actions that modify application state, including requests sent in the background without an open tab. Postman also uses Amazon Bedrock Guardrails to redact personally identifiable information before it reaches the underlying model, with an option for enterprise admins to enable the setting in Agent Mode. The system uses Amazon Bedrock cross-Region inference, with geographic or global profiles available depending on throughput and geographic-processing requirements. Postman says zero data retention is configured for supported models, but availability and behavior vary by model.
Why This Matters
The account makes several production trade-offs visible: limiting the tools an agent sees for a request, supplying context shaped for the task, and retaining controls around actions and sensitive data. For teams assessing developer agents, approval flows, personal-information redaction, and model-dependent data-retention behavior are part of deployment architecture—not just questions of model capability.
Evidence
Start with the primary evidence, then review supporting sources and context.
Primary Evidence
1How Postman runs Agent Mode for 40 million developers on Amazon Bedrock
Srinivas Kini
“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.”
Recommended signals
Recommended by brand, topic, and recent related coverage