Case study
AI Assistant for Contracts Intelligence and Analysis
Building an AI assistant to analyze contract interconnections and augment it with market data
The Client required an advanced AI assistant for their application. The goal was to create a tool that would allow users to answer complex questions about the relationships between contracts and forecast prices, and market trends.
A key technical challenge was the use of a particular LLM model, which lacks native support in standard cloud services like AWS Bedrock, necessitating the development of a custom architectural solution.
The project was carried out iteratively, allowing for flexible adaptation to requirements and systematic value creation. The development process consisted of several key stages:
The project is currently in its final phase, which includes exposing the FastAPI interface and integrating it with the target application in the development environment.
An advanced AI assistant that meets all the client's objectives was successfully designed and implemented. The solution is based on a custom Multi-Model RAG architecture implemented in AWS Lambda (Python), which effectively integrates the LLM model with various data sources.
The architecture utilizes a "Super Agent" with full access to the PostgreSQL database and specialized agents with limited access, ensuring security and operational precision. Additionally, integration with Google Maps allows for the generation of interactive maps, and the analysis of PDF and other documents enriches the query context.
