Project Andrea

Team Info

Team 5 - Freedom

Inspiration

Small Claims cases can be confusing for self-represented users. The information exists, but knowing what matters, what evidence to prepare, and what to do next is not always obvious.

We wanted to use AI not to give legal answers, but to guide users through that uncertainty.

What it does

Andrea helps users:

• check if their dispute may fit the Small Claims Tribunals
• explain what happened in plain language
• organise evidence and build a timeline
• identify missing or unsupported information
• prepare a clearer case summary and next-step checklist

How we built it

Andrea uses a guided conversation instead of an open-ended chatbot. It asks focused follow-up questions, gathers evidence, and grounds responses using structured SCT reference material.

The aim is to combine the flexibility of AI with the structure of a form.

Challenges

The main challenge was balancing helpfulness with responsibility.

We wanted the AI to assist without blindly agreeing with the user, inventing missing details, or presenting uncertain information as fact.

What we learned

Good legal AI is not just about giving better answers. Often, it is about asking better questions.

We also found that responsible AI needs to be built into the workflow and UX, not added as a disclaimer at the end.

What's next

We plan to improve evidence-to-claim traceability, expand grounding material, and make case preparation even more structured and transparent.

Built With

• flue
• react
• typescript