ESG-GenAI: Personalized Investment Advisory Focused on ESG Practices Using Generative AI

Investigates the integration of small and large language models (SLMs and LLMs) to support the generation of accurate, reliable, and personalized financial analyses aligned with ESG principles.

Personalized Investment Advisory Centered on ESG Practices Using Generative AI is a research project focused on the integration of Small Language Models (SLMs) and Large Language Models (LLMs) to support the generation of accurate, reliable, and personalized financial analyses aligned with Environmental, Social, and Governance (ESG) principles.

As financial markets increasingly incorporate ESG considerations into investment decisions, the demand for reliable, personalized, and context-aware financial analyses continues to grow. This project investigates how SLMs and LLMs can be efficiently adapted and combined to assist financial experts in producing high-quality reports grounded in relevant financial data. By leveraging the complementary strengths of smaller models, which enable deployment in secure and controlled environments, and larger models, which offer advanced language generation and reasoning capabilities, the project aims to develop robust and computationally efficient solutions that minimize hallucinations, ensure data privacy, and support informed investment decisions.

Objectives

  • Investigate and evaluate the capabilities of SLMs and LLMs for financial text generation and analysis in Portuguese.
  • Develop efficient adaptation techniques to improve the domain-specific performance and contextual accuracy of smaller language models.
  • Explore strategies for integrating SLMs and LLMs to generate coherent, factually grounded, and personalized financial reports while minimizing hallucinations.
  • Design orchestration and information retrieval mechanisms to extract, aggregate, and incorporate relevant financial and ESG data into generated analyses.
  • Evaluate the computational efficiency, reliability, and sustainability of the proposed solutions, emphasizing secure deployment and responsible AI practices.

Funding & Support

This project is financed by ICT Itaú, covering the period from January 2025 to December 2025.