TricIA: A Multi-Agent System to Facilitate Interaction with Business Intelligence Dashboards

• Bruno Raviolo Universidade de Fortaleza (UNIFOR) http://orcid.org/0000-0001-9080-7702
• Joan Lucas M. S. Chaves Universidade de Fortaleza (UNIFOR)
• Raimir Holanda Filho Universidade de Fortaleza (UNIFOR) https://orcid.org/0000-0002-5754-1372
• Carlos Caminha Universidade de Fortaleza (UNIFOR) / Universidade Federal do Ceará (UFC)

DOI: https://doi.org/10.5753/sbbd_estendido.2025.247658

Resumo

This paper presents TricIA, a multi-agent system based on an open-source language model with 12 billion parameters, designed to enhance the usability of the Monitor Fiscal — a business intelligence platform from the Court of Accounts of the State of Ceará. TricIA acts as a chatbot that assists users in navigating and understanding the available data, both through natural language responses and by performing direct actions on the interface, such as page redirections and filter applications. To ensure efficiency even on modest infrastructures, an architecture composed of three specialized agents was adopted: the orchestrator, the guardrail, and the function-calling agent. This structure distributes responsibilities and allows smaller models to overcome capacity limitations. The paper describes the architecture, integration techniques with the dashboards, and presents a functional demonstration of the system, highlighting its potential to democratize access to complex fiscal information.

Palavras-chave: Multi-agent system, open-source language model, function calling, business intelligence dashboard

Referências

Almeida, F. C. and Caminha, C. (2024). Evaluation of entry-level open-source large language models for information extraction from digitized documents. In Symposium on Knowledge Discovery, Mining and Learning (KDMiLe), pages 25–32. SBC.

Bastos, Z., Freitas, J. D., Franco, J. W., and Caminha, C. (2025). Prompt-driven time series forecasting with large language models. In Proceedings of the 27th International Conference on Enterprise Information Systems, pages 309–316.

Basu, K., Abdelaziz, I., Kate, K., Agarwal, M., Crouse, M., Rizk, Y., Bradford, K., Munawar, A., Kumaravel, S., Goyal, S., Wang, X., Lastras, L. A., and Kapanipathi, P. (2025). Nestful: A benchmark for evaluating llms on nested sequences of api calls.

Karl, A. L., Fernandes, G. S., Pires, L. A., Serpa, Y. R., and Caminha, C. (2024). Synthetic ai data pipeline for domain-specific speech-to-text solutions. In Simpósio Brasileiro de Tecnologia da Informação e da Linguagem Humana (STIL), pages 37–47. SBC.

Lu, Q., Zhu, L., Xu, X., Xing, Z., Harrer, S., and Whittle, J. (2024). Towards responsible generative ai: A reference architecture for designing foundation model based agents. In 2024 IEEE 21st International Conference on Software Architecture Companion (ICSA-C), pages 119–126.

Qin, Y., Hu, S., Lin, Y., Chen, W., Ding, N., Cui, G., Zeng, Z., Zhou, X., Huang, Y., Xiao, C., Han, C., Fung, Y. R., Su, Y., Wang, H., Qian, C., Tian, R., Zhu, K., Liang, S., Shen, X., Xu, B., Zhang, Z., Ye, Y., Li, B., Tang, Z., Yi, J., Zhu, Y., Dai, Z., Yan, L., Cong, X., Lu, Y., Zhao, W., Huang, Y., Yan, J., Han, X., Sun, X., Li, D., Phang, J., Yang, C., Wu, T., Ji, H., Li, G., Liu, Z., and Sun, M. (2024). Tool learning with foundation models. arXiv preprint arXiv:2304.08354.

Tran, K.-T., Dao, D., Nguyen, M.-D., Pham, Q.-V., O’Sullivan, B., and Nguyen, H. D. (2025). Multi-agent collaboration mechanisms: A survey of llms.

Tribunal de Contas do Estado do Ceará (2025). Monitor fiscal - tce-ce. [link]. Acesso em: 02 jun. 2025.