Atticus

Introduction

Atticus helps you transform chaos into clarity. All it takes is a click of a button to harness the processing power of a legally-informed, dynamic AI model at your fingertips. We take raw inputs — messy emails, call notes, texts, or scattered client updates — and turn them into structured case intelligence. Atticus identifies deadlines, missing records, and next steps, linking every detail to the right matter or specialist automatically. Instead of digging through threads or spreadsheets, your team gets instant, actionable summaries — ready for scheduling, drafting, or follow-up. Atticus doesn’t just organize your information; it understands it, so you can move from reaction to resolution faster than ever.

Inspiration

We felt literarily inspired and had begun our application with the character Atticus from the book "to Kill a Mockingbird" in mind. Atticus, being an enigmatic focal point of the novel, seemed perfect for an application that would assist with legal counsel and client updates in a law firm. In fact, the color palette of our site takes inspiration from the atmosphere of the book as well.

How we built it

We utilized Firebase for the database, where we stored both our AI generated tasks and appointments. The Gemini API takes in the user-supplied input and parse it into JSON objects that would serve as the outline for the tasks. Simultaneously, appointments track the date and time, summary, and other features such as the id of the user who booked them. For authentication, we integrated AWS Cognito, importing its login and sign-up components directly into our homepage. Once a user signs in, their Cognito ID token is securely attached to each created task, ensuring proper user association and data integrity across the application.

Challenges we ran into

While we have learned a lot in the last 36 hours about SDKs, generative AI prompting, and even voice bots, we had also run into many hurdles. One of the biggest issues we encountered was with version control within git. During the project we each completed our own tasks and components but when it came time to push them we had to take time to solve merge conflicts.

Another issue we had to tackle was with not defining the database schema ahead of time, and we ended creating more categories than we needed, which made things confusing later on so we had to end up deleting the extra categories. Additionally when utilizing Cognito we initially has issues with storing the tokens.

Accomplishments that we're proud of

We're really ecstatic about how this application turned out under such a tight time constraint. We had been able to successfully connect to both an AWS cloud service (Cognito) and the Google Gemini API, as well as create 4 dynamic AI specialists that demonstrate performing the functions. A huge part of this had been prompt engineering, which had taken some getting used to, but very quickly became intuitive and was one of our biggest assets alongside the voice bot which was able to connect to two of our services.

What we learned

Through this project, we learned the importance of maintaining separate Git branches to prevent merge conflicts and keep our workflow organized. We also realized how essential clear communication is within a team, as there were times when we forgot to update each other on what we were working on, which led to overlapping efforts. Lastly, we discovered the value of taking breaks. When we found ourselves stuck on a problem, stepping away for a bit often gave us a fresh perspective and helped us find a solution more efficiently.

What's next for Atticus

We aspire to further develop this application to encompass more AI specialists and take in input from files, audio, video, etc instead of just raw input. Having learned what we did about React, Vite, TypeScript, Cognito, Firebase, and version control, we could definitely add more onto this application.