Project overview
The backend of a chat assistant that uses knowledge base materials to prepare answers and retains the context of the current conversation.
The FINKO documentation describes a RAG approach: the assistant retrieves knowledge base materials and uses the information found in its answers. This connects the conversation to the content of linked documents.
The backend uses FastAPI with OpenAI integration. The README lists session memory and four language variants: Russian, Uzbek in Latin and Cyrillic scripts, and English. Docker is provided for deployment.
Features and modules
Dialogue: Backend API for the chat assistant.
Knowledge: Search of connected materials through a Vector Store.
Context: Current session memory.
Languages: Russian, Uzbek in two scripts and English.
Integrations
The available materials do not disclose the external connections; describing the integrations requires project documentation.
Artificial intelligence: AI at the core of the assistant
Described in documentation
FINKO uses a language model together with knowledge search. AI is the core of the product, while its sources and memory provide context for responses.
RAG-based search of the knowledge base.
Generating an answer in the context of the current session.
Dialogue in multiple languages.
About this description
Based on the project README. The operational stage and measured results are not confirmed.