AI is ready. Your APIs probably aren’t.

Postman achieved the AWS AI Competency in Agentic AI Tools, and that says as much about your APIs as it does about ours.

I’ve spent a lot of time this year talking with enterprise customers and partners about AI. What strikes me is how consistent the pattern is.

The model isn’t the hard part. The challenge is everything that comes after the prompt.

Organizations can already reach powerful foundation models from multiple providers. The trouble starts when those models need to interact with real systems, business processes, and production data. In June 2025, Gartner predicted that over 40% of agentic AI projects will be canceled by the end of 2027, citing escalating costs, unclear business value, and inadequate risk controls. Gartner’s analysts note that many of these projects are early-stage experiments driven by hype and aimed at problems the technology isn’t suited to solve. They also estimate that of the thousands of vendors claiming agentic AI capabilities, only around 130 represent genuinely agentic products.

The model selection problem is largely solved. The infrastructure problem is not.

Getting APIs ready for how agents actually discover, understand, test, and call them reliably at scale is where most AI initiatives hit friction. Add governance, compliance, and regulatory requirements for enterprises in regulated industries, and the challenge compounds.

That’s why I’m glad to share that Postman has achieved the AWS AI Competency in the Agentic AI Tools category.

What the AWS AI Competency actually validates

The AWS AI Competency isn’t a self-certification. AWS introduced the Agentic AI Tools category to recognize platforms and development tools that help teams build, deploy, and manage agentic AI systems while meeting enterprise requirements for governance, security, and compliance. Partners go through technical validation and have to demonstrate successful customer implementations against AWS’s standards for security, reliability, and operational excellence.

The Agentic AI Tools designation covers platforms that support the full development lifecycle, from design and specification through testing, governance, and production operations, across different regulatory environments.

This isn’t recognition for a demo environment. It reflects the capabilities enterprises need to move AI from experimentation to production and to build the engineering discipline required to operate AI systems at scale.

AI agents run on APIs, and API quality determines agent reliability

The question most enterprises are asking today isn’t whether to use AI. It’s whether their infrastructure is ready for it.

AI agents depend on APIs to reach systems, retrieve information, run workflows, and take action. But there’s a specificity problem that doesn’t exist in human-driven API consumption: a developer can infer intent from ambiguous documentation and adapt. An agent can’t. If an OpenAPI specification omits authentication scopes, misrepresents response schemas, or fails to document error codes and rate-limit behavior, the agent fails. Silently or loudly, but it fails.

As agents become a distinct buyer class that discovers services, evaluates options, and completes transactions on its own, the quality, discoverability, and governance of an organization’s APIs will directly determine whether that organization can take part in an agent-driven economy. An API that a human developer can work through with effort isn’t necessarily an API that an agent can call reliably at scale.

Even the most advanced models can’t compensate for unreliable API foundations.

Postman sits at this layer of the agentic AI stack. By helping teams design well-specified APIs, enforce design consistency through Postman API Governance rules powered by Spectral, and validate API behavior with automated test suites before an agent ever calls an endpoint, Postman lets organizations build AI systems on a foundation they can trust.

And for teams that want to build agents themselves, the Postman AI Agent Builder lets them turn any existing collection into a production-ready MCP server. Existing API documentation, request examples, and tests become the foundation for agent integration. The APIs a team has already built and documented in Postman become tools an agent can discover and call, without a separate integration layer.

Building agent-ready APIs at scale: the PayPal example

We’re already seeing this shift across our customer base.

PayPal shows what happens when an organization treats its APIs as strategic infrastructure built to be consumed by both human developers and AI agents.

Since publishing their public Postman workspace, PayPal’s collections have accumulated more than 100,000 forks, which puts them among the most-forked on the Postman Public API Network. The business impact is concrete: time to first API call dropped from 60 minutes to 1 minute, and testing cycles that used to take hours now finish in minutes. Engineers on the Postman Enterprise plan save roughly one hour per week on API development workflows. The collections are structured, versioned, and machine-readable by design, which is exactly what makes them usable by AI agents as well as human developers.

That agent-readiness is now operational. PayPal published its MCP server as a Postman Collection, presented at POST/CON 2025, giving developers a ready-to-use, fully documented set of API requests that cover payments, invoices, disputes, shipment tracking, subscriptions, and more. The collection uses Postman’s built-in OAuth support, which reduces the authentication friction that typically slows agent integration. The result: an AI agent can discover, authenticate, and call PayPal’s commerce APIs using the same collection infrastructure that human developers already rely on.

Mark Lummus, Head of Product, Developer Tools at PayPal, explains the principle behind it:

“Postman has become a front door to PayPal, and increasingly the developer walking through it is an AI agent. We built our collections and Flows so humans and agents read them the same way, turning discovery into a first API call in under a minute and, over three years, more than 100,000 forks. The AWS AI Competency reflects the platform discipline behind making PayPal agent-ready by default.”

Bringing agentic AI to AWS builders

As part of our growing collaboration with AWS, Postman is helping teams build, test, and govern AI-powered applications directly within the tools they already use.

Postman MCP server in Kiro

Kiro is AWS’s spec-driven agentic IDE, built on the Code OSS platform and designed around structured, intentional development rather than open-ended code generation. When a developer describes requirements in Kiro, it produces three structured artifacts before writing a line of code: requirements.md (user stories and acceptance criteria), design.md (technical architecture), and tasks.md (an implementation checklist). This spec-first workflow gives AI agents and tools exactly the kind of structured context they need to operate reliably.

Kiro includes native MCP support, so it can connect to any MCP-compatible tool server. Through the Postman MCP server, developers can reach Postman workspaces, collections, environments, and tests from inside Kiro without context-switching. Teams can search private APIs in their organization’s catalog, generate client code from existing OpenAPI specifications, run test suites, and keep collections synchronized with their codebase, all within the environment where the code is being written.

Postman Agent Mode on Amazon Bedrock

Postman Agent Mode, our AI assistant, is powered by Anthropic’s Claude models through Amazon Bedrock.

For enterprises, the infrastructure choice matters. Amazon Bedrock offers two inference profiles to match different compliance requirements: In-Region keeps requests within a single AWS region for strict data residency compliance, and Geo Cross-Region routes within a geography (US, EU, and others) for higher throughput while respecting regional boundaries. This architecture lets enterprises use AI-powered developer tooling while keeping the compliance, security, and data residency controls their regulated environments require.

In practice, Agent Mode helps developers generate test suites from existing collection structure, troubleshoot failing requests using workspace context, and keep Postman Collections synchronized with Git repositories as APIs evolve.

API Catalog integration with Amazon API Gateway

The Postman API Catalog integrates with Amazon API Gateway to close the gap between governance and day-to-day API development, and the integration works in both directions.

Teams can import OpenAPI 3.0 and 3.1 specifications (in YAML or JSON) from API Gateway directly into the Postman API Catalog, where they appear alongside associated environment metadata and auto-generated collections. They can view deployment status and CloudWatch metrics within the Postman context. And they can export API specifications back to API Gateway, or deploy HTTP API schemas directly from Postman to a specific API Gateway stage, which creates a continuous loop between design, testing, and deployment rather than a one-way handoff.

The API Catalog is available to Postman Enterprise customers as part of large-scale API governance programs.

Moving AI from pilot to production

The Postman MCP server is available in Kiro today. Agent Mode on Amazon Bedrock is live. The API Catalog integration with Amazon API Gateway is available now.

For AWS customers, systems integrators, and technology partners building enterprise AI solutions, this milestone is about more than a competency badge.

Gartner’s 40% cancellation prediction isn’t a verdict on AI itself. It’s a warning about the gap between experimentation and production-readiness. The APIs, governance, testing, and operational foundations underneath AI systems are what decide whether those systems perform reliably at scale. Getting that infrastructure right is not a model problem. It’s an engineering problem.

As organizations move from AI experimentation to enterprise-scale deployment, preparing APIs for agents will become just as important as selecting the model itself. For many enterprises, it will prove to be the harder challenge.

That’s exactly where Postman is helping teams succeed.

Learn more

• Postman’s AWS AI Competency announcement
• Postman on AWS Marketplace
• The PayPal customer story
• Postman Partners