AI-powered citizen feedback analysis for public service improvement

The problem

Brazil's Federal Executive Branch Public Ombudsman System receives large volumes of citizen feedback, including data from public service user satisfaction surveys. The Ombudsperson's Office monitored its internal workflows and operational performance through predominantly quantitative indicators. While these were useful for tracking operational activity, they did not capture the qualitative dimension of citizens' experiences with public services. The Office needed a way to process large volumes of citizen feedback data at scale, enabling qualitative analysis and generating strategic insights from public service user satisfaction surveys.

The solution

The initiative enabled the rapid, structured, and evidence-based identification of the key factors influencing citizens' experiences with public services. As part of the solution, the Ombudsperson's Office used Gemini 3 Flash to support the processing and qualitative analysis of unstructured comments collected through the Fala.BR platform.

The tool was applied in three main ways: first, to generate a word cloud and identify concepts with higher semantic density; second, to conduct sentiment analysis, classifying comments as positive, negative, or neutral; and third, to cluster comments by recurring themes, such as systems, deadlines, staff approach, and bureaucracy.

By leveraging AI, it became possible to move beyond the predominantly quantitative indicators traditionally used to monitor the Ombudsperson's Office's internal workflows and operational performance, and to incorporate qualitative analyses that reveal users' perceptions, expectations, and opportunities for improvement.

This AI-supported approach helped identify key points of friction, such as perceptions of slow or overly standardised responses, as well as positive feedback related to the Office's agility, effectiveness, and quality of service.

By transforming large volumes of unstructured feedback into actionable insights, the project enhanced the government's ability to understand citizens' needs, strengthened evidence-based decision-making processes, and contributed to the continuous improvement of public services. This project costs approximately $20 a month.

The impact

The project strengthened evidence-based decision-making among the technical teams responsible for public service delivery and contributed to the development of more responsive, accessible, and citizen-centred public services.

1. AI transformed unstructured feedback into actionable insights

The most successful aspect of the initiative was its ability to transform large volumes of unstructured citizen feedback into actionable insights for public management. The use of AI enabled the identification of patterns, perceptions, and opportunities for improvement that would have been difficult to capture through quantitative indicators alone, significantly enhancing the Ombudsperson's analytical capabilities.

2. Specific points of friction and strength were identified across citizen feedback

The AI-supported analysis helped identify the main points of friction reported by users, including difficulties accessing Gov.br services, perceptions of slow or overly standardised responses, and demand for clearer, more personalised communication. It also helped highlight positive perceptions related to the Office's agility, effectiveness, and quality of service.

3. Illustrative examples selected to ground the findings

Gemini was used to select examples of the five most positive and five most critical comments from the satisfaction surveys, in order to illustrate the findings. The analysis showed that positive feedback often associated the Office's work with efficiency, effectiveness, and trust in public services. Critical comments pointed to feelings of invisibility, lack of dialogue, and frustration with technological barriers or generic responses.

Lessons learned

• The solution could evolve towards predictive monitoring. As an area for improvement, the team believes the solution could evolve to incorporate more advanced trend detection and predictive monitoring capabilities. Currently, the analysis primarily helps in understanding past events and supporting corrective actions. In the future, the goal is to strengthen the ability to anticipate emerging issues, enabling a more preventive and proactive approach to improving public services.

• Early investment in predictive tools would strengthen the approach from the start. If starting the project today, the team would invest from the outset in early trend detection mechanisms, automated alerts, and predictive analytics, further strengthening the strategic use of data to support decision-making.

Launch year: 2025

The project was developed within the regulatory framework of Brazil's Federal Executive Branch Public Ombudsman System, which is coordinated by the Office of the Federal Ombudsman-General, part of the Office of the Comptroller General of the Union (CGU). The Federal Ombudsman-General is responsible for establishing the guidelines, standards, and procedures that govern public ombudsman activities across the federal government. The initiative operates within the legal framework established by Brazil's Access to Information Law (Law No. 12,527/2011), the Code for the Protection and Defence of Public Service Users (Law No. 13,460/2017), and other regulations governing transparency, citizen participation, public accountability, and the handling of citizen feedback. The use of Artificial Intelligence was designed to support, rather than replace, human decision-making, ensuring compliance with principles of accountability, transparency, and responsible public-sector AI adoption.