Hi everyone, hope you are all doing well. Today I am going to share how I built an AI agent server and deployed it as API. I'm pretty sure, many of you know by now, what an AI agent is and how much popular they have become.
This is Hack Agno an AI agent which is deployed as an API service.
With the integrations and applications being developed, I also wanted to create such an application. The inspiration came from Youtube when I was, searching for resource materials to learn Agents. later on I saw a clickbait, where a guy once developed and created an Agent deployed as API.
Just around that time, the opportunity presented itself, and then I was already working on building the server. Since this was my first solid AI agent project. I wanted to build with a tech stack that I have worked with before and one I had been working with a long time. So it was Python and FastAPI
The hackathon was focused on building AI agents with any framework so with Python in mind, I opted to use a light-weight AI framework. which is, Agno a lightweight library for building Agents with memory, knowledge, tools and reasoning.
You can use Agno to build Reasoning Agents, Multimodal Agents, Teams of Agents and Agentic Workflows. Agno also provides a beautiful UI to chat with your Agents and tools to monitor and evaluate their performance.
I created different API routes, that mostly are powered by AI agents. In this project, there are two AI agents, which basically communicates with FastAPI backend, then the response is generated and the output is then show at the client side.
I also had free time so I decided to give it a retro terminal like UI, and am really satisfied with it.
Check out the Original server Here
For those who want to see this in action can go to
By building this project is taught me a lot of things, also showed me that we can use AI agent in many different themes and setups to automate, improve and make apps way more better and interactive.
If you like this project You can vote for it. Login to the Quira and vote on the Hack Agent Server
Follow for more such content.
Top comments (7)
Idk if you are interested but there is a fastapi library that will automatically take fastapi endpoints and convert them to mcp for llm tool calls. You could extend this as a tool function to another LLM via mcp. It might not make sense but each tool can be exposed independently for easy extension.
thx for this @mr_bloodrune Will try doing that
Ok let's go
I would like to dig it more and understand further. Tknaks!
Sure, You can get the link from the project vote page
very interesting
Nice work! Love how clean and modular the setup is — especially the focus on making agents stateful. Looking forward to seeing how you expand this further!