LawLinker

• 

  Front Page

• 

  The User Interface where the clients can enter their stories and create a case quickly and easily

• 

  A diagram of our project

• 

  Data that we will have for our future machine learning implementation

Inspiration

After our discussions with Morgan and Morgan, we identified the hurdles in the client onboarding process. It was evident that the traditional form-filling experience was cumbersome. Driven by the idea of enhancing user experience, we combined modern technology and intuitive design. Our solution doesn't just cater to those who may not speak English but streamlines the entire process for all, making legal assistance accessible and hassle-free!

What it does

Our project revolutionizes the client onboarding process for law firms like Morgan and Morgan. By harnessing advanced AI technology, we've eliminated the tediousness of form-filling. Clients simply voice their case details, and our system, leveraging OpenAI's models, captures, translates, and organizes this information into a structured database. This not only simplifies the client's experience but also provides attorneys with immediate, comprehensive case insights, making legal assistance more efficient and user-friendly.

How we built it

We're no strangers to OpenAI's GPT, but the Whisper model was new terrain for us. Integrating this audio-to-text feature was a game-changer. To bring it all together, we relied on the MERN stack - a powerful combination of MongoDB, Express.js, React, and Node.js. This blend allowed us to craft a seamless application that integrates the prowess of AI with robust web development tools, offering a streamlined experience from the backend to the frontend.

Challenges we ran into

• First time using the Whisper AI model or any audio-to-text technology, it was hard learning to use it in a short time.
• Correct prompt engineer for GPT to grab a client's story and efficiently transform that information into an efficient data structure.
• Prompting GPT to create our own data-set so we could analyze and make statistics about cases.
• OpenAI policies making GPT be against lending aid towards legal cases and legal advice, we had to make sure our prompts wouldn't trigger the safety mechanisms that shut down GPT from replying.

Accomplishments that we're proud of

Whisper takes any language and correctly transcribes information in English', even if the client speaks multiple languages at the same time! Our CRUD operations work efficiently, the client just needs to tell their story and once its complete, its submitted immediately! We used this absolutely new technology and were able to produce a working MVP within 36 hours!

What we learned

We learned alot about how does audio-to-text technologies work, AI prompt engineering, doing CRUD operations in a non-relational database like MongoDB and trying our best to efficiently manage our project and what features can we realistically add.

What's next for LawLinker

While working on the project, we discovered a realm of new possibilities. We want to place the data into a dataset, to extract key information such as the most common type of incidents, and the most common locations of incidents. Another key functionality that should be implemented in the future, is a way for the clients to choose their attorney, based upon their expertise. Functionality such as a rough compensation estimate, or a virtual FAQ guide could be implemented by looking at previous records of similar cases and after a discussion with an attorney.

Built With

• express.js
• gpt-4
• javascript
• mongodb
• node.js
• openai
• react
• whisper