BotanicalAI

• 

  Landing page

• 

  Chatbot with domain knowledge

• 

  Users upload their data

• 

  Analyze custom data

Inspiration

New immigrants are struggling to find reliable agents for purchasing homes. We aim to completely transform this industry.

What it does

Our software utilizes a combination of public sources, partner data sources, and our own proprietary data to train the LLM model. This enables realtors to analyze data and provide personalized recommendations to meet the unique needs of their customers.

How we built it

For full-stack web development, we utilize Django. To provide data to the LLM, we use llama_index. Additionally, to connect with OpenAI chatGPT, we employ lang_chain.

Challenges we ran into

Our team initially grappled with limited expertise in Django and a lack of familiarity with llama_index. However, through dedicated time investment in poring over documentation and persistent application, we triumphed over these challenges to expand our knowledge and skills.

Accomplishments that we're proud of

We successfully harnessed the power of chatGPT-4, enriching it through meticulous fine-tuning with a fusion of public data and our synthetic dataset.

What we learned

We've learned to utilize the new technology to speed up our software development project and create a fine-tuned model.

What's next for BotanicalAI

We will finish the database part (potentially with Pinecorn) to store our data and model. Next, we will deploy it to Azure Cloud service, and complete the core features before releasing.

Built With

• chatgpt
• django
• langchain
• llama-index
• openai
• python