How Axians powers faster and smarter document summarization with AI

Handling large volumes of technical narratives is a daily reality for organizations managing complex public programs. Within the Programa Demografia, Qualificações e Inclusão (PDQI), teams rely heavily on detailed analyses, technical opinions, and structured content within the SI PESSOAS2030 platform to support critical decision-making processes. Transforming this information into concise, consistent summaries is essential to ensure efficiency and clarity across workflows.

In this context, PDQI partnered with Axians to integrate the Document Extraction Retrieval Assistant (DERA) into its operations, enabling intelligent text summarization that streamlines repetitive tasks, improves consistency, and enhances the overall user experience. This collaboration marks an important step toward more efficient and standardized processing within the PESSOAS 2030 program.

Susana Nunes, Head of Unit of the Systems Information and Innovation Unit at PDQI, shares her perspective on the challenges, benefits, and future potential of AI-powered summarization within the organization.

What is your organization’s activity and what role does text summarization play in your day-to-day operations? 

Our organization manages complex operational and analysis processes within the PESSOAS 2030 program, namely across application analysis, payment request analysis, and related technical workflows supported by the SI PESSOAS platform. In this context, documents, technical opinions, and structured narrative fields are part of daily work, and text summarization plays an important role in helping teams turn detailed technical content into concise, consistent outputs that can be used more efficiently in the decision-making process.

In practice, summarization is especially relevant in the “Descrição de Resultados” context, where the goal is to support the automatic completion of a results description based on the technical rationale already written by the analyst. This helps reduce manual drafting effort, improve consistency and quality of the text produced, and minimize omissions or formulation issues in repetitive tasks.

Before implementing Axians’ DERA, what challenges were you facing in reading, interpreting, or analyzing the summarized text?

Before this type of capability was introduced, the work depended much more on manually reading the technical rationale and then rewriting or condensing that information into a concise and operationally useful summary field. In a context where teams deal with many records and recurring procedural texts, that created a heavier manual burden and made it harder to maintain the same level of consistency across outputs.

As with any AI-enabled summarization feature, another challenge was ensuring that the generated wording stayed fully aligned with business requirements and with what the technical teams wanted to communicate. Internal exchanges show the importance of keeping the output faithful to defined requirements, avoiding unintended wording changes, and refining the model when needed.

How does DERA support your teams in summarizing long texts, and what impact have you seen in terms of time savings, efficiency, or reduced repetitive work?

DERA supports our teams by helping transform longer technical narratives into shorter operational text that can be reused directly in the SI PESSOAS workflow, particularly in the “Descrição de Resultados” process. From a user perspective, its main value is not only speed, but also standardization: it reduces repetitive drafting work, supports more uniform wording, and helps analysts focus more on validating content than on rewriting it from scratch.

The impact we highlight most is qualitative: less manual effort in repetitive summarization tasks, more consistency in the resulting text, and a more streamlined user experience inside the business process. We are also treating this as an evolving capability, continuously refining prompts, expected output patterns, and monitoring how often users accept or adjust the generated text to improve the experience further.

Over the past year, since its implementation in March 2025, DERA has helped our teams process 8582 summarization tasks, of which 5693 were not changed (66%) and 2889 were changed (34%), contributing to estimated time savings of 429 hours and allowing users to focus more on validation and decision-making rather than repetitive drafting.

How was the collaboration with Axians during the implementation of DERA?

The collaboration with Axians was close, practical, and iterative. The implementation involved alignment meetings, functional presentations, feedback loops with business users, and follow-up adjustments whenever the output needed refinement. This was especially important to ensure that the summarization feature matched the real wording and operational expectations of the teams using it daily.

We would describe the collaboration as constructive and responsive: Axians remained available to discuss requirements, present the functionality, correct issues, and adapt the model when feedback from the business side showed that changes were needed. That iterative approach was essential to turning the feature into something useful in a real operational setting rather than just a technical demonstration.

By integrating DERA into its workflows, Programa Demografia, Qualificações e Inclusão (PDQI) has streamlined document summarization, reduced repetitive manual effort, and improved consistency across thousands of technical analyses. The collaboration with Axians and the continuous refinement of AI-generated outputs have helped teams save time, enhance productivity, and focus more on validation and decision-making.