Automating Institutional Internal Communication with a Generative Artificial Intelligence Agent and Scheduled Workflows at IFTO - Araguatins Campus
Internal communication in public educational institutions is often hindered by administrative overload, fragmented formal channels and slow information dissemination. This article describes the development, deployment and extension of an internal communication agent based on generative artificial intelligence at the Federal Institute of Tocantins (IFTO) - Araguatins Campus. The research, applied in nature and qualitative-descriptive in approach, followed an evolutionary prototyping cycle conducted at the Artificial Intelligence and Automation Research Laboratory (LEIAA). The solution, with a hybrid architecture (self-hosted Docker containers and cloud services), combines the n8n orchestration platform, the PostgreSQL database, the Evolution API messaging gateway and the GPT-4o-mini language model, integrated with Google Workspace services through the Model Context Protocol (MCP). The system operates in two modes: in the reactive mode, the language model interprets natural language requests and performs calendar, e-mail and spreadsheet actions; in the proactive mode, scheduled routines query institutional spreadsheets and send messages built from templates. Between May and June 2026, five proactive workflows (institutional events, undergraduate thesis defences, birthdays, academic calendar and scheduled messages) operated in three real WhatsApp groups of the campus, comprising staff, students and faculty. During this period, notices were sent daily according to the spreadsheet records, and the team observed no failures in the event and birthday dispatches. A comparison with systems described in the literature indicates that the solution combines actions executed by a language model, proactive messaging to groups and self-hosted orchestration; its effect on the communication team's routine, however, has not yet been measured.
Authors
- Fábio Emanuell Abreu Cardoso (ORCID: https://orcid.org/0009-0005-8785-2058)
- Nicolas Heitor Feitosa Costa (ORCID: https://orcid.org/0009-0001-8240-3123)
- Ramásio Ferreira de Melo (ORCID: https://orcid.org/0000-0001-9847-3460)
- Valéria dos Santos Silva Torres (ORCID: https://orcid.org/0009-0002-1994-0230)
- Gabriel Ribeiro Milhomem Ferreira (ORCID: https://orcid.org/0009-0006-0135-5235)
- Allan Carlos Araújo Teixeira (ORCID: https://orcid.org/0009-0007-9563-7969)
- Julio Cesar Santos Sousa (ORCID: https://orcid.org/0009-0002-3478-6585)
Institutions
- Instituto Federal do Tocantins (BR)
Publication Details
- Journal
- Revista Multidisciplinar do Nordeste Mineiro
- Published
- 2026-11-16
- DOI
- https://doi.org/10.66104/2fxv0d97
- Primary Topic
- Robotic Process Automation Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00