Inspiration
We want to utilize the vector database and llm with bring your own model to make the ecosystem more affordable and we want to focus more on features and retrievers with agents support.
What it does
We want to make a more affordable and enterprise-ready website voice
- chatbot that allows users to bring their own model, such as Pinecone or OpenAI. Our platform will focus on advanced features and enhancements, such as:
- Chat widget control
- Memory management
- Embedding chunk control
- Search retriever (Langchain/Llama index)
- Agents support (Zapier/API/SQL DB)
How we built it
Pinecone, OpenAI, Langchain, Llama Index, Vercel, Digital Ocean
Challenges we ran into
Hallucination, Prompt control, and ensuring our chatbot low code solution is compatible with upcoming changes to support langchain agents / openai function calls.
Accomplishments that we're proud of
Bring your own model helps user to come up with their own pinecone. Customizaiton of LLM, Embedding, chunk control., retrievers and langchain agents makes our app unique in the market for startups to enterprises.
What we learned
A long way journey to utilize langchain, llm, pinecone to get better accuracy.
What's next for Talkingsite - Website Chat + Voice Bot
Ecom search, agent with API and SQL connect with the customer backend application. Chrome extension to support for social media reply, social media reply and social media chat.
Log in or sign up for Devpost to join the conversation.