I recently participated in HackTrinity as part of Team Neural Litigators.
We built a legal AI assistant.
Presenting to you, The Lawful Llama 🦙!
The Pitch
Imagine you’re a tech lawyer working with multiple businesses looking to innovate in the Gen-AI space in the European Union(EU) or businesses that handle large amounts of data.
You may be working with multiple clients at the same time. Hence, you would also need to spend a LOT of time researching the ever-evolving legal-tech landscape of the EU.
The EU has been a pioneer in AI and Data Protection regulations, and the stakes are high for all of your clients if they don’t comply with these laws, leading to fines worth millions of Euros.
Here’s what your research process might look like before you give out any legal advice:
You would need to go over 100+ pages of documents involving laws like the GDPR and the EU AI Act.
Stay up-to-date with the landmark judgments passed in the context of the GDPR and the EU AI Act.
You would want to read articles from reputed sources to understand the interpretations of the laws before making any advice.
Lastly, thoroughly review your client’s policies to confirm compliance.
As a tech lawyer, wouldn’t you wish for a tool that could possibly automate this tedious research process so that you could spend more time forming some quality advice for your client?
Well, I present to you The Lawful Llama - A RAG-enabled Legal AI Research Tool that would assist every tech lawyer in their research process, making them more productive and enabling them to take up more clients simultaneously!
The Product
Here’s a preview of the website.
The Lawful Llama is a RAG-enabled AI chatbot specializing in the EU’s take on AI and Data privacy regulations.
It has the knowledge of over a million words of documents like:
GDPR (General Data Protection Regulation)
The EU AI Act
White papers from the EU Parliament
Case Reports from the EU Court of Justice
Cases involving the Data Protection Commission of Ireland
Interpretations of laws and cases by reputed Law Firms.
Let’s see a demo of how it works.
Demo
You will see the following interface when you click the “Try the Chatbot” button.
Imagine if you’re a lawyer at a banking firm looking to build some Gen-AI products, and you want an exhaustive overview of compliance, risks, and further readings that you may need to go over.
This works how every chatbot works, attach your company documents, enter your query, and get your answer generated by AI.
Here’s what a typical response of our chatbot would look like.
How Is It Different?
We kept asking ourselves this question during the development process of our product. How is it any different than the LLMs that we have right now?
And the answer is simple: Context.
Our legal chatbot has unparalleled, up-to-date context that no other chatbot has regarding the EU’s AI and Data Privacy regulations.
It utilizes RAG to extract the most relevant content for a particular query a lawyer enters. It gives a detailed answer, including an analysis of compliance, potential risks your client may face, and suggested readings for the lawyer to review.
Moreover, it has the ability to search the internet if the database doesn’t have the relevant context. All thanks to Perplexity API!
For instance, if a client’s documents aren’t fully compliant with the GDPR, our chatbot would suggest reviewing the Meta Ireland vs. Data Protection of Ireland 2023 judgment.
It’s all a game of context and precision.
My Experience At HackTrinity
It’s been around a year since I’ve come to Ireland, and this was my 2nd hackathon in a year. Read my experience from my very first hackathon in Ireland here!
And I really don’t have much to say, except that it was great!
I was lucky enough to get teamed up with great people. Shout-out to them for building this project without any significant hassle!
We split the work into maintaining the front and back end for this project and achieved most of what we planned to do.
We were satisfied with what we built even though we didn’t win. Plus, free food for the weekend!
Final Notes
I’d like to extend my thanks to the team at HackTrinity for organizing this hackathon and I would also like to thank my teammates for working on this project with me.
Lastly, if you’ve read till here, I recommend you to subscribe to my newsletter!
I’ll post part two of this blog, explaining this project’s architecture and backend!
So make sure you subscribe to the newsletter so you can read my blog straight from your inbox.
You’ve reached the end of this blog. Thanks for reading!