Financial education is identified as a challenge in Brazil. In recent years, it has been integrated transversally into the school curriculum, meaning it is taught among traditional subjects such as math and Portuguese. The promotion of this knowledge occurs in a context of increasing the number of investors. The proportion of Brazilians who invest in some financial product increased from 31% of the population in 2021 to 37% in 2023, according to Anbima (Brazilian Association of Financial and Capital Market Entities).
To facilitate this learning, technology can be an ally. With this in mind, students Arthur Motta Pereira Gomes, Fernando Vieira dos Santos, Gabriel Zahar Mine, Leonardo da França Moura de Andrade, and Sergio Eduardo Ramella Junior, under the guidance of Professor Graziela Tonin, completed a Capstone project (the final project for engineering programs) focused on implementing improvements to B3's financial education ChatGPT.
This chatbot, available on the "Bora Investir" portal, aims to assist investors through exclusive content from B3 and CVM (Securities and Exchange Commission of Brazil). "The challenge presented by B3 to us was the low audience engagement with this chatbot, featuring short-duration interactions and minimal user retention, which is why the students worked on enhancing the tool," highlights Graziela. According to the professor, this lack of engagement compromises the chatbot's effectiveness, making it difficult to create a solid connection with investors and limiting the tool's impact in disseminating financial knowledge.
B3's chatbot is based on OpenAI's artificial intelligence language model, one of the leaders in this technology. Companies have been applying AI to revolutionize customer support, content creation, and data analysis with the generation of insights. B3's tool responds to users' financial market inquiries impartially. "This means ensuring that the technology responds reliably and free from conflicts of interest. Additionally, this interaction with users needs to be as understandable as possible through simple and clear language," observes Graziela.
Among the improvements made by the students are creating mockups of the application screens, developing a quiz on financial knowledge, formulating layout and functionality recommendations to enhance user experience, and prototyping these improvements in a programming environment. "All these steps were carried out through a validated prototype that followed an agile methodology, with frequent structured deliveries and constant validations with the B3 team, where the feedback collection allowed the prototype to evolve and consequently improve the end-user experience in sprints," explains the advisor.
The quiz aims to increase engagement through a fun experience that practically tests and reinforces user knowledge, showing at the end whether the answers are correct or not. This tool is part of a central effort of the Capstone to enrich B3's interaction with the audience through aspects such as customization.
Furthermore, the work done by the Insper students implemented a model that facilitates the integration of the chatbot with B3's platforms, such as the B3 Education portals, B3 Investor, and the HUB3 app. This model was developed as a standalone page, not attached to any platform, so it can be accessed from any of B3's platforms. Another advantage is that it was built on a microservices architecture, facilitating its integration with the mentioned portals. "This expands the chatbot's reach and ensures that the user experience is personalized and adapted to different audience needs and profiles, providing accessibility and flexibility across platforms," says Fernando.
According to the student, the Capstone experience has been interesting from the beginning, especially in implementing this solution for a real client, including weekly validations. "Being in an environment like B3 was challenging and helped to understand how the work dynamics in a large company operate," he adds.