Generative artificial intelligence-assisted automation of bulletin publishing on a university institutional blog

Abstract

A university unit sustains the periodic circulation of specialized analytical content through an institutional blog hosted on a widely used blog publishing platform. This platform operated with an outdated publishing interface and relied on the manual, individual upload of each item of content, resulting in a backlog of more than 100 items pending publication. Based on the unit's own favorable prior technical experience using generative artificial intelligence, this type of support was adopted as guidance to build an automated solution scheme.

Development began with the support of one generative artificial intelligence model, which was replaced, following an initial evaluation, by a second model; the latter was used to structure an architecture organized into three components: one focused on data, another on visual presentation, and another on executing the publication process. Based on this scheme, the work was carried out in five successive stages: the automated retrieval of data from documents stored in the cloud, the design of an editorial template with institutional identity, the configuration of a secure connection between the data source and the destination platform, the execution of the publication process, and the adjustment of the site's overall visual identity.

As a result, the entire accumulated backlog was processed, and the previous interface was replaced with one featuring institutional identity, a functional preview, and direct document download, while also demonstrating the capacity to sustain a stable editorial flow in subsequent periods. The experience made it possible to acquire and consolidate competencies in cloud-based authentication, execution quota management, and service integration without relying on traditional development frameworks.