Landing Page
New chat
Speech to text
Chat
Procedures
Actual Files
Steps

Inspiration

When we saw the "Accessible Digital Government" challenge, it instantly resonated with our team, the Cachirulitos. We've all experienced — or seen our parents experience — the absolute headache of trying to figure out which documents are needed for a simple government procedure. We realized that the barrier isn't just a lack of digital portals; it's a lack of accessible language.

We were inspired to build a bridge between citizens and their government, transforming intimidating bureaucratic portals into friendly, conversational experiences.

What it does

GuiaMX is an intelligent, conversational agent designed to facilitate access to public procedures.

Users simply type what they need to do, for example:

"I need to register my new small business"

GuiaMX responds with a highly accessible, easy-to-understand checklist and guide. It strips away the complex legal terminology and replaces it with plain, clear language, pointing users exactly where they need to go without the usual confusion.

How we built it

Our development process was a journey of iteration:

Ideation & Restructuring

We started with a very broad idea. We quickly realized we needed to narrow our focus, so we completely restructured our concept.

Requirements Gathering

Once we had a solid, focused idea, we mapped out the exact system requirements, focusing heavily on user experience and accessible language constraints.

Development Foundation

We built the core conversational architecture using Microsoft Copilot Studio, which allowed us to structure the dialogue flows effectively, and Azure Speech for speech recognition capabilities.

AI Integration

To give GuiaMX its "brain" and the ability to dynamically simplify complex government jargon, we integrated Azure OpenAI as the powerhouse behind the chatbot. Azure Speech was used for Speech Recognition.

Challenges we ran into

We jumped into the deep end with this hackathon! Our biggest hurdles were entirely technical:

Zero Prior Agent Experience

None of us had ever built an AI agent before. Figuring out how dialogue nodes, intents, and variable passing worked took a lot of trial and error.

Learning Azure OpenAI on the Fly

We had absolutely no prior experience with Azure OpenAI. Navigating the documentation, understanding the API connections, and figuring out how to prompt the model effectively to guarantee accessible language — all while the clock was ticking — was incredibly stressful but rewarding.

Pivoting Early

Scrapping our first idea to restructure and write solid system requirements felt like we were losing time, but it ended up saving the project.

Accomplishments that we're proud of

Going from Zero to One

We are incredibly proud that we managed to build a fully functional AI agent from scratch, having never touched someof these technologies before.

Mastering New Tech

Successfully connecting Copilot Studio with Azure OpenAI and watching the chatbot respond accurately for the first time was a massive win for the team.

Designing for Real Impact

We built something that isn't just technically cool, but genuinely addresses a defined social problem. GuiaMX has the potential to actually help people in our communities.

What we learned

The Power of Planning

We learned that stepping back to write solid system requirements and restructure an idea is vastly more efficient than blindly writing code.

Microsoft Ecosystem

We gained hands-on, practical knowledge of how to orchestrate intelligent workflows using Microsoft Copilot Studio and Azure OpenAI.

Prompt Engineering for Empathy

We learned how to carefully prompt an AI to adjust its tone, ensuring the output is not just accurate, but genuinely empathetic and easy to understand for everyday users.