The OECD.AI Policy Navigator

Our policy navigator is a living repository from more than 80 jurisdictions and organisations. Use the filters to browse initiatives and find what you are looking for.

Use of Artificial Intelligence Models in the Qualitative Evaluation of Responses - Avalia


Added by:   OECD analyst
Added on:   30 Sep 2026
Updated by:   OECD analyst
Updated on:   30 Sep 2026

Anatel has developed AvalIA, a tool that uses artificial intelligence models to automatically assess the quality of telecommunications companies' responses to consumer complaints. This initiative seeks to make customer service more efficient and transparent, benefiting both consumers and the regulatory agency itself. AvalIA was created to expand and improve the analysis of companies' responses, ensuring that consumers receive more appropriate information and more effective solutions.

Name in original language

Uso de Modelos de Inteligência Artificial na Avaliação Qualitativa de Respostas - Avalia

Initiative overview

The National Telecommunications Agency (Anatel) has sought to improve the relationship between consumers and regulated service providers, using advanced technologies such as Artificial Intelligence (AI) to analyse and evaluate the handling of consumer complaints through the Anatel Consumidor system. The main issue that the AvalIA system seeks to resolve is the need for a more assertive, efficient, and transparent diagnosis of how companies respond to consumer complaints, ensuring compliance with regulatory standards and promoting information quality in the sector. In addition, the initiative aims to take advantage of the opportunity to automate processes that currently depend on manual human analysis, making monitoring more robust and comprehensive.

The objectives of the AvalIA initiative are clear and specific: to automate the process of Qualitative Response Assessment (AQR) by providers to consumer complaints, estimate the subjects of complaints, identify the enabled information indicators, and verify that these indicators have been properly addressed. This is expected to improve the effectiveness of resolving demands, increase transparency, and encourage continuous improvement in service delivery through each provider's Information Quality Index (IQI). The tool also ensures the anonymisation of personal data, guaranteeing the protection of sensitive consumer information.

The expectation is that AvalIA initiative will evolve into a fully implemented and institutionalised system within Anatel, enabling the automated assessment of all complaints registered in the Anatel Consumidor system. This will allow for the inclusion of a larger number of providers in the process, faster assessments and more timely diagnoses, contributing to a more efficient and transparent regulatory environment. The tool can be expanded to cover other types of demands and indicators, as well as adapted to other complaint databases in the regulated sector, maximising its impact and reach.

In summary, AvalIA represents a significant innovation in the way Anatel monitors and evaluates providers' response to consumer demands. By incorporating AI into the qualitative assessment process, the Agency strengthens the diagnosis of information quality, ensures compliance with consumer rights,
and promotes the continuous improvement of services provided, consolidating itself as a benchmark for transparency and regulatory efficiency in the telecommunications sector.

Other relevant details

Results, outcomes and impacts: AvalIA enables automated assessment of 100% of complaints registered with Anatel Consumidor, expanding the scope of the Qualitative Assessment of Responses (AQR) and generating significant gains in efficiency, consistency of regulatory criteria and timeliness of diagnostics, as well as increasing the number of participating providers. Its performance is supported by robust metrics, based on a history of 143,000 manually evaluated responses used to validate the AI models. Going forward, faster and more accurate detection of systemic informational failures in providers’ responses is expected, improving demand resolvability and strengthening responsive regulation for consumers who turn to Anatel and those who seek to solve problems directly. Challenges and lessons learned: During AvalIA’s development, key challenges included ensuring data quality and representativeness for model training, coping with time‑varying data, and handling the complexity and volume of around 110,000 complaints per month, with unbalanced classes and multiple subjects per demand. These were addressed through continuous data updating and validation, systematic monitoring of model performance with appropriate metrics, and careful operational integration with the Anatel Consumidor system, which required technical adjustments and close alignment among multidisciplinary teams. Lessons learned highlight the need for high‑quality, sufficiently large datasets, periodic model updates to follow evolving demand patterns, and the recognition that state‑of‑the‑art technology is not always the most suitable option in a regulatory context. Success factors include strong institutional commitment, collaboration between domain experts and technical teams and clear objectives.

About the policy initiative


Category:

  • AI policy initiatives, programmes and projects

Initiative type:

  • AI use cases/projects in the public sector

Status:

  • Active

Start Year:

  • 2025

Target Sectors:


OECD AI Principles:

—

Other relevant urls:

—