Negotiator Genius

Negotiator Genius Logo

Inspiration

For the Gen Lab's AI Agents and Fintech Hackathon, the category that appealed to us the most was autonomous systems to transact on behalf of their users. Some of the members of our team are entrepreneurs who spend time and money negotiating contracts and would like an AI Agent to do all this work for them.

What it does

We created an AI system that understands legal contracts and negotiates with the counterparty on your behalf.

How we built it

On the backend we used Python, LangChain, and Chat GPT 4. On the front end we used Gradio and Jupyter.

Challenges we ran into

The biggest challenge we faced was figuring out the reasoning of the AI Agent to understand a negotiation conversation. We extracted relevant terms from the contract then hacked the prompts and with each negotiable term the AI Agent was able to negotiate with the counter-party (human).

In the beginning we were challenged in creating the right scope for the project. The UI Design was also challenging because we had to simplify our project to make it achievable.

Accomplishments that we're proud of

We are proud of successfully using prompt engineering and LangChain to extract and negotiate terms in a contract.

Translating legal subject matter expertise in a form that can be used in prompt engineering is another accomplishment.

What we learned

We learned how to build our team and communicate as a team and build a successful project together. We were able to manage our time successfully to finish a working prototype.

What's next for Negotiator Genius

We created an architecture that can be iterated quickly, which did not contain a database. We would like to add a database like NoSQL to store documents and current and past negotiations.

We would also like to use Negotiator Genius to monitor market conditions and amend contracts to reflect optimal pricing.