Postman has just raised the stakes in the AI development ecosystem with a major unveiling at POST/CON 2025, its annual developer conference. Already the go-to platform for over 40 million developers worldwide, Postman is now positioning itself as the central workspace for building and scaling intelligent systems powered by APIs. The company introduced a comprehensive suite of capabilities tailored specifically for developers grappling with the challenges of building AI agents—those complex, often brittle pieces of software that sit atop large language models and must interact robustly with a web of services.
At the core of this evolution is Postman Insights, now in open beta. It offers real-time observability into API usage, bringing visibility into how both humans and agents interact with endpoints. This visibility extends to version tracking and failure detection—essential for teams running intelligent systems in production, where an unexpected shift in a payload or a subtle authentication error can cause agents to silently break. The new Repro Mode takes this a step further, letting developers debug API failures using real-world request data, which is especially useful when testing agents that operate autonomously with unpredictable behavior.
The data backing Postman’s move is compelling. Between May 2023 and April 2025, Postman saw a surge in traffic from AI model APIs. Meta’s Llama usage grew sevenfold, Google’s Gemini quintupled, and lesser-known but increasingly flexible models like Replicate and Mistral saw their traffic double and quadruple respectively. Yet despite the diversification, OpenAI still commands 60% of total AI-related traffic—a reminder of its staying power even as alternatives gain traction. This fragmentation across model APIs underscores why developers need a toolset that can abstract, test, and monitor across a heterogeneous model landscape. Postman is betting that its platform will become the connective tissue.
The introduction of the Model Context Protocol (MCP) into Postman’s workflow marks a significant advancement. This protocol standardizes how agents call external APIs and brings structure to what has until now been an improvisational process. With the Agentic AI Builder and native MCP integration, developers can turn any API into an MCP-compliant tool, generate a server from a Postman collection, and simulate interactions with AI agents—all inside Postman. Furthermore, the newly announced MCP Server Network allows developers to discover trusted endpoints at scale, verified and curated for production use. It’s the beginning of a formal registry for agent-safe APIs, which could prove to be a vital layer in an increasingly autonomous web.
Beyond the core agent features, Postman has deepened its enterprise integrations. New workflows with GitHub, Jira, Slack, and Microsoft Teams are designed to break down silos and accelerate iteration cycles. These integrations allow for branch-specific governance, seamless syncing, and real-time issue creation—all of which are foundational for companies deploying APIs at enterprise velocity. The introduction of Notebooks further reduces onboarding friction by merging interactive documentation with live API calls—turning a static spec into an executable tutorial.
With over 100,000 APIs and 18,000 publishers on its network, Postman’s infrastructure is already vast. But what’s different now is its clear alignment with the next frontier of software: intelligent agents. As noted by CEO Abhinav Asthana, this isn’t just about smarter prompts—it’s about building dependable, testable systems governed with the same discipline as any other production software. Agent systems, if left unchecked, can hallucinate, fail silently, or introduce unanticipated risks. The architecture needed to contain that complexity is exactly what Postman aims to provide.
Industry leaders are already taking notice. Alex Chriss of PayPal praised the platform for driving not just speed, but developer velocity—a subtle but crucial distinction that points to the value of sustained, high-quality iteration. In the emerging era of agentic commerce, velocity is everything. The faster developers can test and scale AI interactions across an ecosystem of APIs, the more agile and impactful their services become.
As the lines blur between developer tools and AI infrastructure, Postman is positioning itself not just as an API platform, but as a command center for building the intelligent systems of tomorrow. Whether this vision holds in the long-term will depend on adoption and trust—but with this rollout, Postman has offered developers the scaffolding to move beyond experiments and into fully operational agentic architectures.