Name in original language
Mentor.IA
Initiative overview
MentorIA is a conversational virtual assistant developed by the Secretariat of Management and Innovation of the Ministry of Management and Innovation in Public Services (Seges/MGI), in partnership with Serpro. Integrated into the Compras.gov.br system, it uses generative artificial intelligence with RAG (Retrieval-Augmented Generation) technology to answer, in real time, the questions of procurement officers based exclusively on current official documents.
The system reduces the time spent searching for regulations, avoids rework, supports decision-making, and strengthens the transparency and legality of government procurement processes. The initiative is aligned with the Brazilian Artificial Intelligence Plan (PBIA) and its strategic axis No. 3, AI for Improvement of Public Services. It also operates in the public procurement dimension, aiming to incorporate AI tools into public administration routines to increase the efficiency and quality of government decisions.
Unlike open AI tools available on the internet, MentorIA provides traceable responses, indicating the regulatory sources and the exact excerpt that supported each answer, ensuring greater reliability and legal certainty for procurement processes.
Launched in March 2026 and available to all Compras.gov.br users, MentorIA is in its initial phase, with a focus on procurement planning. The next steps include deeper integration with procurement systems, expansion of the indexed document base, and extension of its scope to other phases of the public procurement cycle.
The initiative represents a concrete step in the modernisation of Brazilian public management, contributing to more efficient procurement and higher-quality services for society.
Results, outcomes and impacts
Launched in March 2026, MentorIA serves all users of Compras.gov.br, the procurement portal used by civil servants across hundreds of Brazilian federal agencies. As the system is recent, quantitative results are still being consolidated.
During the acceptance testing phase in December 2025, the system was validated for response accuracy based on indexed official regulatory documents. Quality is continuously monitored through user feedback using thumbs-up/thumbs-down ratings and audit logs.
Expected impacts include a significant reduction in the time spent on regulatory consultation, a lower incidence of bid challenges, and greater legal compliance in the planning of public procurement processes.



























