Inspiration
As a computer science engineering student balancing coursework, exams, and side projects, I found myself constantly context-switching to check on simple things—like whether my latest front-end push built correctly, or what my schedule looked like for the day. I wanted a hands-free way to interface with the tools I already use without breaking my flow state while coding.
What it does
It acts as a seamless bridge between the Amazon ecosystem and my daily development workflow. Instead of opening a dashboard, I can simply ask for an update. It retrieves the latest data points from my custom backend and reads them out in a clean, straightforward summary.
How we built it
I kept the architecture strictly problem-first. I set up an open Model Context Protocol (MCP) server to handle the logic, connecting it directly to my existing backend APIs. I purposely avoided over-engineering it with heavy frameworks, relying instead on lightweight scripts and AWS for fast, reliable routing.
Challenges we ran into
Getting the payload sizes right for the voice responses took some trial and error. I initially tried to pull too much raw data at once, but quickly realized that retrieving thousands of data points within a strict 4-kilobyte payload limit is physically impossible. I had to rewrite my queries to fetch and format only the essential, actionable insights.
What's next for DeployBot by ORION
I plan to refine the custom authentication flow so multiple users can safely query their own private dashboards on the same device. For now, it is a lightweight, effective prototype that genuinely saves me a few minutes every day.
