CivicFix

Inspiration

Our original idea started with a specific problem: immigrants and exchange students often have to navigate unfamiliar government processes and documents without knowing where to start.

We realized that this problem extends far beyond newcomers. Students, families, immigrants, and everyday citizens regularly have to deal with government forms, applications, benefits, taxes, permits, and other official documents that can be difficult to understand.

Government information is publicly available, but finding the right information, understanding what applies to you, and figuring out what to do next can still be a major challenge.

We wanted to build something that could bridge that gap: a tool that helps people navigate government processes in a way that is simple, accessible, and grounded in official information.

What it does

Our project is designed to help people navigate government documents, processes, and services without having to search through dozens of government webpages themselves.

Instead of being limited to one government service, our goal is to create a flexible system that can support different use cases. Our current implementation focuses on building a structured knowledge base around Ontario student tax information, using official CRA and Ontario government sources as our starting point.

The system tracks authoritative government sources through a Source Registry, allowing information to be organized while maintaining a connection to the original source.

This creates the foundation for a future assistant that can:

Help users understand government documents and requirements.
Identify which government information is relevant to a user's situation.
Break complicated processes into clearer steps.
Direct users to the appropriate official resources.
Provide answers that are grounded in verified government sources rather than relying solely on a general-purpose LLM.

Our vision is to make navigating government services feel less like searching through a maze of webpages and more like having a clear guide through the process.

How we built it

We built the foundation of our system using Make for workflow automation and Google Sheets as our knowledge store.

Rather than simply connecting an LLM to the internet and asking it to answer government-related questions, we focused on building a structured source layer first.

Our workflow:

Identify authoritative government sources relevant to a specific use case.
Register those sources in our Source Registry.
Organize information into a structured knowledge base.
Automate the collection and processing of information using Make.
Maintain source traceability so information can be connected back to the official government resource.

Our student tax use case serves as the first implementation of this broader approach. The same architecture can be expanded to other government services and document-heavy processes.

Challenges we ran into

Our biggest challenge was figuring out what the actual workflow should look like.

It was easy to say that we wanted to build an AI assistant that helps people with government documents. The harder question was figuring out how we would actually get reliable government data into the system, how we would organize it, how we would keep it updated, and how our solution would provide something meaningfully different from simply asking an existing LLM the same question.

We had to think carefully about:

Where our data would come from.
Which sources we could trust.
How we would structure government information.
How information would flow through our automation.
How we could maintain the knowledge base as government information changes.
How to make our system source-grounded and traceable rather than just another general-purpose chatbot.

We also faced technical limitations with our available data-storage tools. Instead of letting those limitations stop development, we adapted our architecture and used Google Sheets as our knowledge store while continuing to build the automation in Make.

Accomplishments that we're proud of

We are proud that we moved beyond the idea of simply building another chatbot and began developing the information infrastructure behind a trustworthy government assistance tool.

Some of our key accomplishments include:

Building a Source Registry for authoritative government resources.
Creating an automated workflow using Make.
Establishing Google Sheets as our structured knowledge store.
Building our first use case around Ontario student tax information.
Designing the system around official government sources and source traceability.
Creating an architecture that can eventually be expanded beyond taxes to other government services and document-related processes.

Most importantly, we created a foundation that can grow into something much broader than our initial use case.

What we learned

One of our biggest lessons was that building a useful AI solution is not just about the model itself. The information pipeline behind the model is just as important.

We learned that a system can only be as useful as the information it has access to and the way that information is organized.

Throughout the project, we learned how to:

Identify and evaluate authoritative government sources.
Structure unorganized information into a usable knowledge base.
Build automated workflows using Make.
Design a source registry for maintaining information.
Think about data freshness, reliability, and traceability.
Adapt our architecture when technical limitations arise.
Think critically about what makes an AI solution different from simply using a general-purpose LLM.

Most importantly, we learned to approach AI development from the perspective of building a reliable system around the model, rather than treating the model as the entire solution.

What's next for CivicFix

Our current student tax use case is only the beginning.

Our next step is to expand CivicFix beyond taxes and into a broader government document and service navigation platform.

We want CivicFix to eventually help users with a variety of government processes by understanding their situation, identifying the relevant government requirements, breaking complicated processes into manageable steps, and directing them to the appropriate official resources.

Future improvements include: