Offline

NitroStack at Humans x Agents Bengaluru

NitroStack joined Humans x Agents in Bengaluru as a sponsor and demonstrated an end-to-end MCP workflow, from building and testing an MCP server to NitroCloud deployment and ChatGPT integration.

Bengaluru, India
NitroStack at Humans x Agents Bengaluru

On May 23, 2026, Project Humans in the Loop and Agentic AI For Good brought together AI builders, founders, engineers, and researchers in Bengaluru for the first Humans x Agents meetup.

NitroStack joined the event as a sponsor, with NitroStack CEO Abhishek Pandit presenting a live end-to-end Model Context Protocol (MCP) demo.

The idea behind the session was simple: instead of spending a talk explaining what MCP could eventually make possible, build something in front of the room and take it all the way to a usable application.

That meant starting with an MCP server, testing it, deploying it, and finally making it available as a ChatGPT App.

All in under 10 minutes.

Humans x Agents brought MCP into a builder-first room

Article image

Humans x Agents was the first meetup organized by Project Humans in the Loop together with Agentic AI For Good. The Bengaluru edition received more than 150 registrations for a room designed for roughly 50 people.

The format leaned heavily toward builders. Short demonstrations covered AI agents, workflows, MCP, developer tooling, and projects people were actively working on rather than long conference-style presentations.

That made it a useful setting for the kind of MCP conversation we wanted to have.

MCP is easy to explain at the protocol level. The harder question for a developer is what happens after that: How do I actually build an MCP server? How do I test it? Where do I deploy it? And how does it finally reach something people can use?

Abhishek's session focused on that entire path.

From a pizza app to a deployed MCP server

For the live demo, Abhishek used a pizza-ordering application as the starting point.

Inside NitroStudio, he demonstrated the vibe-coding workflow for MCP apps, creating the MCP server and moving directly into testing the application inside Studio.

The demo then moved from local development to deployment on NitroCloud.

The flow looked roughly like this:

Idea → MCP app → Test in NitroStudio → Deploy on NitroCloud → ChatGPT App

There wasn't a second demo prepared to represent the production side of the story. It was the same application moving through the complete workflow.

Once the MCP server was live, Abhishek connected it to ChatGPT and showed the pizza application working there as an MCP-powered ChatGPT App.

The complete build-to-ChatGPT path was demonstrated in under 10 minutes.

The interesting part wasn't really the 10 minutes

Speed made the demonstration easy to understand, but it wasn't the main point.

MCP development can quickly become fragmented. A developer might understand how to expose a tool, but still have to piece together scaffolding, testing, debugging, infrastructure, deployment, client integration, and UI before someone can actually use what they've built.

The Humans x Agents demo showed those steps as one workflow instead.

That is also where we want NitroStack to sit for MCP builders: between “I have an idea for an MCP app” and “someone can actually use it.”

NitroStudio gives developers a place to build, inspect, test, and iterate on MCP applications. NitroCloud handles the deployment side. From there, the same MCP server can move into clients such as ChatGPT rather than remaining a local prototype.

Making that path visible was more useful than another slide explaining why MCP matters.

An early step in how NitroStack talks about MCP

Article image

Humans x Agents was a relatively small community event, and that was part of what made the session useful.

For NitroStack, it became an early milestone in taking MCP outside documentation and developer conversations and explaining it through something people could watch being built from beginning to end.

Create the server.

Test it.

Deploy it.

Use it inside ChatGPT.

That sequence made MCP much easier to understand.

As NitroStack continues working with developer communities, workshops, buildathons, and AI events, this is the kind of awareness we want to build: less discussion around MCP as an abstract protocol, and more exposure to what developers can actually ship with it.

For developers who want to try the workflow themselves, the NitroStack documentation covers the MCP development stack in more detail.