Abhinav Asthana

Postman Collections have become the de-facto way in which people document, test, and share APIs with each other. Collections combine executable requests with documentation and tests that allow for anyone to get started with an API easily for new users while also allowing experts to build deep workflows spanning everything from simple tests to integrations and more. Not only use cases but the scale of Postman Collections have expanded too. Collections can contain thousands of API requests, scripts, and reusable components. We have been working on solving for these pain points over the last few months and I am pleased to share that we will be introducing a new Collections v3 format soon in the Postman app with several enhancements: 1. Instead of a single JSON, the format is now powered by YAML that allows for easy diffing and review within git. 2. Breaking down the single view into multiple component files and folders that allows for granular modifications by humans as well as AI agents. 3. Multi-protocol support for REST, gRPC, GraphQL, MQTT and many other formats in a single representation. Building on top of our recently announce native git integration, we believe this will help improve developer workflows. Check out the video for a demo. Release coming in the next few weeks!

There's a new way to work with Webflow in @getpostman via Agent Mode, and it feels like magic 🪄 Add the Webflow MCP server, ask the agent to fetch your data, it works out the right API call and runs it. Save the request to your collection. Done. Learn more in the blog: wfl.io/4uucD8Y

Astro AI is a production-grade control plane for AI agents that grew out of our own experience building real agents in production at Postman. It makes agents discoverable, governable, and deployable as first-class operational assets inside your organization. Three problems kept surfacing that we turned into the foundation of Astro AI. 1️⃣ First, data ingestion and tool connectivity are solved problems that every team shouldn't have to reinvent, so we built reusable, composable primitives for connecting agents to internal systems and external APIs. 2️⃣ Second, knowing what's running, where, and under what constraints should be a given, so we built in consistent observability, guardrails, and deployment operations from the start. 3️⃣ Third, the mechanical work of wiring agents together slows down the part that actually matters, so we structured Astro AI to absorb that burden and let teams focus on experimentation, iteration, and shipping to production faster. The core of Astro AI is an open harness spec: execution graph, tool contracts, guardrail policies, data substrate, observability config, and deployment metadata in a single versioned artifact. A packaged agent is portable across runtimes, model-agnostic, and consumable via HTTP or surfaced directly in Slack. The CLI is available now to start building. The registry is by invitation and we want your input. Check it out at astropods.com.

I have been blown away by the response from customers and developers about the new Postman launch! The new Postman is AI-native and it is reflecting in adoption. We just crossed more than 3 million developers now using AI and Agent Mode in Postman. Not just in terms of usage, the rate of new customers buying our paid plans has grown by more than 500% since launch. More than 2/3rds of our new customer cohorts are using AI features in Postman and the rate is only growing. While we are excited about this, we are working on many exciting capabilities: 1️⃣ Long-running and autonomous modes for agents in Postman 2️⃣ Bringing the right context across thousands of APIs and services and millions of datapoints for the agent to access. 3️⃣ Execution sandboxes for agents to do their work securely. 4️⃣ Simulations to allow agents to know the outside world without touching real APIs. All of this is possible with the work we have done over the last few months with v12 and the new Postman. If you missed the announcement, you can read it here: blog.postman.com/new-postman-is…

Your API work now lives in Git. Specs, tests, mocks, and collections move through branches and pull requests, just like your code. postman.com/new-postman/?u…

Our CEO, @ivanburazin, opening Compute with a look at what’s next for dev environments, AI agents, and the infrastructure powering it all.

Building APIs is critical for survival for the agentic era. Great post by @levie Building APIs that actually work is hard though and can’t be left to engineering alone. It’s as much a business decision as it’s an engineering decision.

The Atlassian Rovo MCP server + @getpostman Agent Mode brings agentic workflows to API development and troubleshooting. With context from the Atlassian Rovo MCP server, developers can use Agent Mode to seamlessly turn API runs into Jira work items, updated specifications, documented decisions in Confluence, and more. For more you can do with the Rovo MCP Server ➡️ go.atlassian.com/py15l6

Many people asked me about my comment around the new Postman being AI-native and what does it mean. While the detailed answer has many parts that we'll publish in the coming days, a big part of it is the Postman CLI. AI Agents are very good at using CLIs and have proven them to be very effective in developer-centric workflows. We are building every new capability in Postman with the CLI in mind. The Postman CLI can run collections, mocks, flows, lint specifications, as well as automate workflows around workspaces. The list is exhaustive and growing: learning.postman.com/latest-v-12/do…

The new @getpostman UI brings everything into a more unified view, letting me organise Specs, Collections and Environments exactly where I need them. I can also customise the sidebar and remove items that aren’t relevant to that Workspace’s context.

loving the new @getpostman v12 so far dude

@yessabhishek Let me know. Many perf improvements are in this release especially in the local dev loop. More on the way.

@a85 Just downloaded the update, seems clean and well organised. will test it for few days before coming on the speed part. Thanks :)

Today we're introducing a new Postman platform built for the agentic era. Similar to what happened a decade ago with the platform shifts of cloud and mobile, AI agents are driving the platform shift today. Agentic systems take probabilistic actions at runtime and also consume more API calls than traditional software. APIs are the authoritative interface between agents and the real world. Getting them right has never mattered more. The new Postman is built for that reality: Git-native workflows, an API Catalog that serves as a live system of record across your entire organization, Agent Mode (native AI that works end-to-end across your Postman assets and connected repos), and a full-stack of capabilities to help teams take their APIs from dev to prod for internal, external, and agentic use. This is the most significant platform evolution we've shipped. I wrote about what this all means for our users here: blog.postman.com/new-postman-is… Welcome to the new Postman.

羽田に戻ってきた。今回のSF出張の主な目的はSKO参加だったが、サブ目的として@nagixと一緒に書いたAPIファースト本(翔泳社)をPostmanファウンダ 3人に直接渡すがあった。結果、うち2人（Abhinav、Ankit）に手渡しできたのは、本当に嬉しい瞬間だったぜ✌️

@calcsam That said, we like Mastra a lot. Team is building agents with it too.

@calcsam It’s not the same space. Personally, I have lost count of the number of frontend JS framework projects. Also, it looks like recently companies are all vibe coding each other’s projects too so I expect more pressure on names there. Don’t know whats going to survive eventually.

Last week, we introduced Astro AI — our platform for discovering, managing, and operating AI agents. Today, we’re shipping the first major milestone! This release brings the core infrastructure needed to move agents beyond experiments and into production. To make this possible, Astro AI introduces a declarative spec that defines an agent’s topology — models, knowledge stores, integrations, ingestion pipelines — separately from how and where it runs. With this release, you can: 🔸 Define agents using a declarative spec 🔸 Package them into portable, versioned artifacts 🔸 Include data, memory, tool contracts, and deployment metadata 🔸 Build and push agents via CLI AI agents shouldn’t be scattered prompts and scripts. They should be structured, versioned, observable, and owned. You can install the CLI and run agents locally today: docs.astropods.ai This marks an important step toward agent-native infrastructure teams can trust in production. Interested in early access for running these in production? Sign up for the waitlist: blog.astropods.ai/waitlist

@calcsam Thanks Sam! Continuing a bit with the space theme here.

@a85 looks like a cool product! curious about the naming choice

If agents start consuming developer infra at scale -- hosting, DBs, domains -- seems like winners will be services built for API-first signup, orchestration, and payments. No browser, no human in the loop.

The new Claude Sonnet 4.6 is now live on @getpostman Agent Mode! Anthropic's newest Sonnet model brings major upgrades to planning, reasoning, and code generation tasks.