Visivel as an epistemological infrastructure for territorial social assistance: Humanized registration, georeferenced evidence, and responsible predictive governance

This article aims to examine Visivel as an epistemological infrastructure for territorial social assistance aimed at people experiencing homelessness. Rather than treating the platform only as administrative software, the study analyzes how it produces, protects, and organizes public knowledge about a population whose presence is often fragmented across streets, services, reports, and institutional records. The methodological corpus combines the public interface available at the Visivel web system, the implementation guide, and the technical resource report prepared for public managers. The analysis identifies four epistemic operations: humanized identification, georeferenced territorialization, intersectoral interpretation, and responsible predictive governance. The article argues that Visivel's central contribution lies in converting situated encounters and citizen reports into protected traces that can support planning, outreach routes, service referrals, monitoring, and accountability under SUAS, the National Policy for the Homeless Population and the LGPD. Its predictive layer is interpreted as a decision-support mechanism, not as an automated decision-maker, because institutional responsibility remains with human managers and social workers. The proposed model contributes to debates on data-driven social policy by showing that technical sophistication in vulnerable contexts depends on privacy, territorial accountability, and epistemic humility.