Postman Launches Orbit to Help AI Agents Find the Right API Fast

Quick Reads

Postman has launched Orbit, a free API discovery tool built for AI agents
Orbit gives agents two tools: search to find the right endpoint and integrate to generate step-by-step integration briefs
It is available as an MCP server and a REST API, with zero sign-up or authentication required
Orbit uses up to 27x less context than reading full API documentation
It is built on top of the Postman API Network and is live at buildwithorbit.ai

Postman has launched Orbit, a free Postman Orbit API discovery tool designed to help AI agents find and integrate the right public endpoints, without any setup or authentication. The announcement, published on August 27, 2026, comes from Postman co-founder and CEO Abhinav Asthana.

The problem Orbit solves is straightforward. AI agents get their capabilities from APIs. However, finding the right API and integrating it correctly on the first attempt has remained a consistent challenge. Orbit closes that gap.

The tool gives agents two distinct operations. First, search accepts a plain-language task description and returns a ranked list of matching public endpoints, each graded on fit, including what it explicitly cannot do. Second, integrate accepts selected endpoints and returns a task brief: the auth scheme, base URL, numbered request steps, expected response codes, and inter-step dependencies.

According to the official Postman blog post, splitting discovery into two turns is deliberate. Choosing the right endpoint is precisely where an agent working from memory gets it wrong, and it’s the decision worth keeping in human hands. postman

Orbit is available in two forms. Developers working inside a coding agent such as Claude Code or Cursor can connect it as an MCP server with a single command. Those building their own agent or backend can, instead, call two REST endpoints directly at api.buildwithorbit.ai. Both options expose the same functionality, with no API key required.

The efficiency gains are significant. Based on figures from the Postman blog, an Orbit search and integrate pair for a PayPal invoicing task uses roughly 2,500 tokens. By comparison, reading PayPal’s official OpenAPI specification costs approximately 69,000 tokens, about 27 times more. Additionally, fetching PayPal’s documentation page returns 180KB of HTML, of which 60% is script tags, leaving the agent unable to access the actual parameter-level details it needs.

Speed is also a factor. Across 16 different tasks tested, the median search returned in 7.6 seconds and the median integrate in 9.2 seconds, putting most workflows at 15 to 20 seconds end to end.

A key feature of Orbit is its FIT verdict system. Every task brief opens with a FIT field that states clearly whether the selected endpoints fully, partially, or cannot cover the task. Postman describes this as the feature they care most about, noting that a discovery tool that always finds something, even when it shouldn’t, is worse than useless.

Orbit draws its data from the Postman API Network, the public corpus of real API collections configured and tested by developers. This means the integration briefs it generates reflect actual request schemas, saved response examples, and live auth configurations, not inferred documentation.

For developers whose agents don’t automatically reach for Orbit, Postman has also released an installable Claude Code skill that instructs the agent when and how to use the tool. The skill can be scoped to a single project or applied globally across all Claude Code sessions.

Orbit is live now at buildwithorbit.ai and its full OpenAPI specification is publicly accessible. For Nigerian and African developers building on AI agent infrastructure, Orbit offers a meaningful reduction in both token costs and integration time, two friction points that directly affect the economics of shipping API-powered products in emerging markets.