Postman.
The plumbing
of the internet.

Open any developer's laptop in 2026. The orange icon is already running.

Postman is the workspace where the modern internet gets wired together. Designers sketch endpoints in it. Engineers debug responses in it. Security teams audit collections in it. And, increasingly in 2026, AI agents are reaching into it on their own to chain APIs together without a human moving a cursor.

The company sits at a strange and useful crossroads. It is, technically, a developer tool - which usually means a small but loyal market. It is also, somehow, infrastructure - the connective tissue inside an estimated 98% of the Fortune 500. That combination is the entire story.

A Chrome extension that grew up to run the plumbing of the modern internet. — The short version

APIs were everywhere. Nobody had agreed on how to actually use one.

By 2012 the API economy was real. Stripe was minting developers. Twilio was minting phone calls. AWS was minting nearly everything else. Companies were shipping endpoints faster than they could write documentation for them.

And yet, on the developer's side of the wire, the experience was awkward. Hand-rolled cURL commands. Personal Postel-style tolerance for malformed JSON. Screenshots pasted into Jira tickets. The discipline of building APIs was racing ahead of the etiquette of using them.

The problem wasn't the protocol. HTTP was fine. The problem was the absence of a shared place to talk about a single request - to design it, run it, save it, share it, test it, and stake your name on it. Engineering at scale runs on shared artifacts. The API world didn't have one.

Three engineers from Bangalore bet that the missing piece was a polite, shareable request.

Abhinav Asthana wrote the first version of Postman in 2012 while working at Yahoo's Bangalore office. He needed a less painful way to test APIs at work, so he built one for himself, named it after the HTTP POST verb, and shipped it for free on the Chrome Web Store. Within months it had tens of thousands of users he had never met.

That was the data point. He pulled in two former classmates and colleagues, Ankit Sobti and Abhijit Kane, and in 2014 the three of them turned the side project into Postman, Inc. The bet was unfashionable at the time: a free tool, given to individual developers, with no enterprise sales team and no early revenue plan.

Build the thing you wish existed. Then keep building it for fifteen years. — Founder logic, condensed

One workspace that quietly turned into a platform.

Most tools either solve one problem deeply or many problems shallowly. Postman, somewhat unfashionably, decided to do both. It is a request runner. It is also a designer, a documentation generator, a mock server, a test framework, a monitor, a registry, a publishing surface, and now an AI agent host.

The unit that holds it all together is the Collection - a sharable, executable group of API requests. Collections are why a single developer with a Postman tab open is, in practice, plugged into a network: she can fork a workspace from Salesforce, run it against her staging environment, write tests on top, share the result with a teammate in Berlin, and schedule the whole thing to run inside CI by lunchtime.

What you can actually do with it

Design REST, GraphQL, gRPC, SOAP, WebSocket and MQTT endpoints. Write contract and integration tests, in code or in natural language via Postbot. Generate documentation that doesn't lie because it is executed against the same collection devs use. Mock unfinished APIs so the frontend team isn't blocked. Monitor production endpoints. Publish a public workspace so your developer community can fork it. Build an AI agent that calls your APIs on a customer's behalf.

Postbot, briefly

Postbot is the AI layer that has been creeping further into the product since 2023. It writes tests from a request body. It explains a 502 in plain English. It debugs a failing collection by reading the response. It is, importantly, trained on a corpus very few other people have - billions of real API calls flowing through Postman - which is the kind of moat that does not show up in a marketing deck.

The Collection is the GIF of API development. Small, sharable, instantly understood. — Why this tool spread

The numbers stopped being cute somewhere around the seventh figure.

Postman's adoption curve is unusual because almost none of it was bought. Developers found the extension, kept it open, brought it into their next job, then quietly upgraded their team to a paid plan when the free seat limit got annoying. The growth lived inside individuals' habits before it showed up on a CFO's invoice.

Customers include Shopify, PayPal, Salesforce, Cisco, Intuit, AT&T, the BBC, Stripe and Twilio. The public API Network now lists more than 100,000 published workspaces, which is its own quiet platform business: companies don't just use Postman, they ship through it.

If APIs are the building blocks of software, the workspace is the workbench.

Postman's official story is straightforward: make APIs easier to build, test, document and use. The slightly more interesting version is that the company has bet, repeatedly, that the API itself is the most durable abstraction in software. Languages change. Frameworks die. Cloud providers swap places. The interface, the contract between two systems, persists.

That bet looks different in the agentic era. The new generation of AI agents does not browse the web for fun - it calls APIs. Whoever owns the workspace where APIs are designed, described, governed and tested is, by happy accident, also positioned to govern the substrate that agents run on.

The API was the most underrated abstraction in software. Postman noticed first. — The whole thesis, in one line

The agents are coming. They will not be using your dashboard.

The next category of software won't click buttons. It will read schemas, parse responses, retry on a 429 and chain a dozen endpoints together to satisfy a user request. That category of software is also bad at hallucinating a working request out of thin air. It needs a registry of known, tested, well-described APIs. It needs governance. It needs audit trails. It needs, if we are honest, a workspace.

This is where Postman's last fourteen years - the Collections, the public API Network, the test runners, the governance model, the documentation that actually executes - quietly compound into something more strategic than a tool. The thing developers used to debug endpoints turns out to be the thing AI agents will use to actually do work.

Open the laptop again

Open any developer's laptop in 2026 and the orange icon is still running. Only now, more often than not, something else is using it too: a script, a CI job, a cron, an agent. Postman is no longer just where developers debug APIs. It is where machines start to politely ask each other to do things. Which is, if you think about it, exactly what the postman has always done.

If you'd rather listen.