Kai, our AI chatbot, leverages a combination of Retrieval-Augmented Generation (RAG) and agent mode to deliver accurate and dynamic responses. Here’s how it works:
Document Indexing and Retrieval
Gleap indexes all your documents, websites, and FAQs into a vector database. When a customer asks a question, we search for the closest documents and provide them to the Language Model (LLM) to generate an accurate response. This process involves several steps, including reranking, vector search, and embedding the question.
Integration of Custom Functions
In addition to document indexing, Gleap allows the integration of custom functions or dynamic content through APIs or SDK functions. This information is fetched in real-time and is not stored in any database.
Data Security and Privacy
It's important to note that the data is never used to train our AI models. We only transform the data into a vector representation and store the vector along with the document in a database for retrieval purposes.
We use OpenAI’s LLMs to generate answers. For security and availability, we utilize their Enterprise APIs, ensuring that the data is not used for future training and is deleted immediately after the LLM processes the request.