Postman’s AI Engineer Wows Developers at San Francisco Demo
Eighty developers packed a room in San Francisco last week, and none of them expected one small schema edit to expose ten hidden problems in seconds. Postman’s Agents and APIs meetup at Canopy in San Francisco hosted Talia Kohan, who gave her first public demo of the AI Engineer to a room of about 80 developers. However, the real story is what happened next.
Kohan opened with a bold claim. Coding is easy, she said, but context is not. That single line set the tone for the entire Postman AI Engineer demo. Indeed, every developer in the room understood the pain instantly. After all, code rarely lives in isolation. Instead, a tiny schema change can ripple through APIs, services, and teams that an AI tool never sees.
To prove her point, Kohan made one small edit. She changed the response schema of a single API, adding a new field with a slightly different shape. Then, she asked the AI Engineer a simple question about downstream dependencies. Within minutes, the answer arrived, and it stunned the crowd.
The AI Engineer flagged ten artifacts that needed updates, including five other APIs, the backend models file, the main Flask application file, the test suite, and the documentation. Consequently, the room fell silent. Furthermore, Kohan pushed the point further. If one tiny edit touched ten files, then a full migration or authentication rewrite could quietly break dozens more. Nevertheless, most teams would never catch these breakages until production failed.
So, what makes this Postman AI Engineer demo different from typical coding assistants? The answer lies in the Context Graph. This Context Graph maps relationships across artifacts, giving the AI Engineer a constantly updated picture of how systems connect. As a result, the tool can reason across an entire architecture instead of starting fresh with every prompt. Moreover, it understands which services consume which contracts and which workflows might break.
During the Q&A, one developer asked what actually sits inside the Context Graph. The graph holds APIs and their OpenAPI specifications, databases and services, Postman Collections, monitors, environments, and even third-party tools like Sentry, Datadog, and New Relic. Therefore, nearly every corner of a company’s API ecosystem becomes visible, searchable, and ready for action.
Another developer asked why the team chose Slack as the first launch point. Postman brought the agent to where developers already work, with more invocation points such as the API, Claude, GitHub, and the Postman CLI on the way. Clearly, this approach avoids forcing engineers into yet another browser tab.
For Lagos-based fintech and e-commerce teams juggling fast-moving microservices, this kind of dependency mapping could save hours of manual auditing before every release. Meanwhile, the AI Engineer remains in beta, and any team with an existing Postman workspace can try it today by connecting the Slack integration. Afterward, simply ask the agent to trace a risky change before it ships.
