ToyotaChatbot

Inspiration

We attended Arize AI's workshop yesterday, and it truly inspired us to explore the field of GenerativeAI and RAGs.

What it does

It generates response to the prompts asked by the customer regarding the information of all the Toyota car models. It can give personalized recommendations, compare the features between multiple car models on the basis of existing data.

How we built it

We used LLMs to process data to give personalized recommendations. We employed LLamadeploy and ChatGPT 4.0 mini, on top of the all-cars-dataset filtered out for Toyota cars. We used Reflex for the frontend.

Challenges we ran into

The dataset was very sparse and big compared to what OpenAI supports for Vector Embedding, this make things difficult.

Accomplishments that we're proud of

Successfully deployed Real-Time Agent using state-of-the-art LLM Framework.

What we learned

We learned how to use LLM Frameworks like LangChain, LangFlow, and Llama_deploy to create business-specific ideas and their advantages over LLMs like ChatGPT4, and Llama that work based on knowledge. These tools enable seamless integration with APIs, databases, and custom workflows, making them highly adaptable for domain-specific applications.

What's next for ToyotaChatbot

We can implement an interactive agent that can create accounts for users, removing the processing overhead. Through this, the user can get chat-specific car advertisements during the product's discount periods, and it helps the company advertise targeted offers effectively, improving customer engagement and boosting sales by reaching the right audience at the right time.

Built With

• chatgpt
• llamadeploy
• reflex