Reply Solution
The solution proposed by Technology Reply is based on a generative AI engine implemented in Azure AI Services, designed to extract ESG KPIs from documents in PDF format, which contain sustainability information. The engine uses advanced database embedding techniques through Azure AI Search with semantic engine to handle large volumes of unstructured data and make them available for quick analysis and searches. The architecture is built using the Langchain and Langgraph libraries for workflow orchestration and integration of different components.
The solution leverages language models available in Azure OpenAI Studio, which can be selected and configured to meet specific customer needs, ensuring flexibility and customization. The Retrieval Augmented Generation (RAG) system implemented through Azure AI Search allows you to enrich the model capabilities with domain-specific knowledge, significantly improving the accuracy and relevance of ESG KPI extractions.