[MASTER] Multi-Agent Sales-Tailored Expert Robot

sample
logo

Inspiration

Better awareness about EVs will help reduce gas emissions, which is crucial for the environment.
Also automation of various tasks leads us to better future, where people can only do hobbies.

What it does

Our project is multi-agent bot, that provides high-quality advice for customers interested in EVs, tailoring suggestions to individual profiles and preferences.

How we built it

We utilized a multi-agent architecture along with GPT-4-Turbo and LangChain to develop a robust, conversational AI that integrates with the data from the Mercedes Benz API for up-to-date information and offers.

Challenges we ran into

Ensuring data privacy while personalizing user interactions.
Handling data integration from multiple sources.
Optimizing the AI and multi-agent structure for natural and effective communication without hallucinations.

Accomplishments that we're proud of

Developing a system capable of understanding and interacting with over 9 customer profiles in two major markets (DE and US) with more than a 100 car modifications. Implementing a sophisticated architecture that can manage multiple AI agents to provide cohesive advice after considering all the facts.

What we learned

A LOT about agent coordination and NLP
How to tailor AI-generated advice to diverse customer needs

What's next for [MASTER] Multi-Agent Sales-Tailored Expert Robot

1) Integrate additional data and prompt-tuning methods to refine and personalize customer interactions further.
2) Expand the market scope beyond DE and US to include other regions.
3) Develop more advanced techniques for data retrieval and user profiling to enhance the recommendation accuracy.