---
title: "Cidadão.AI: Multi-Agent AI System for Democratizing Access to Brazilian Government Transparency Data"
description: "Bachelor's thesis (Computer Science, IFSULDEMINAS). Presents the design, development, and evaluation of Cidadão.AI — a platform with 17 specialized agents using NLP, multi-agent architecture, and LLMs to provide conversational access to Brazilian government transparency data. Written in Portuguese."
author: "Anderson Henrique Da Silva"
date: "2025-12-19T00:00:00Z"
updated: "2026-04-04T12:02:48.500663Z"
category: "publication"
tags: ["Design Science Research","Machine Learning","Software Engineering","NLP","Government Transparency","Multi-Agent Systems"]
canonical: "https://www.ntlabs.dev/en/blog/publication-cidadao-ai-monografia"
locale: "en"
---

**DOI:** 10.5281/zenodo.18550748

**Zenodo:** https://zenodo.org/records/18550748

**Type:** Dissertation (Bachelor's thesis)

**Institution:** IFSULDEMINAS — Federal Institute of Education, Science and Technology of Southern Minas Gerais

**Program:** Bachelor of Computer Science

**Defense:** December 2025 — Maximum grade (10/10)

**License:** CC-BY 4.0

**Language:** Portuguese (full text available on Zenodo)

---

This work employs Design Science Research to investigate orchestration patterns for LLM-based autonomous agents in the domain of Brazilian government transparency. The resulting artifact — Cidadão.AI — is a platform with 17 specialized agents (named after Brazilian historical figures) that integrates natural language processing, multi-agent architecture, and large language models to democratize access to public data.

The platform processes over 30 federal and state transparency APIs, performing semantic routing via BERTimbau embeddings, contract classification with fine-tuned models, anomaly detection through FFT spectral analysis, and RAG with tri-layer caching.

**Keywords:** Design Science Research, Machine Learning, Software Engineering, Natural Language Processing