Aim

To design a collaborative platform allowing teachers and their students to create, customize, and converse with educational and cultural characters inspired by Brazilian figures — using generative AI to make schools not passive consumers of technologies, but active participants in their understanding, content development, and experimentation.

Origin of the name

The term Ai_Monguetá is borrowed from the Língua Geral, a hybrid language forged in colonial times from Old Tupi that long served as a means of communication between Indigenous peoples, colonizers, and enslaved Africans. In its broadest sense, ai-monguetá can be translated as "to speak"¹. The word monguetá also evokes, in contemporary Guarani, the idea of "to converse," and takes on a deeper meaning in the Mbyá-Guarani vocabulary: "to inspire" or "to make the word circulate"², valuing orality as a vehicle for transmission and social bonding.

With a contemporary reading and a touch of poetic license, the word also evokes the world of artificial intelligences — the IAs, or AIs in English. The name therefore carries a double gaze: one turned toward Brazilian linguistic and cultural roots, the other toward the transformations that emerge at the encounter between human and digital forms of communication.

¹ A. Gonçalves Dias, Dicionário da Língua Tupy, chamada Língua Geral dos Indígenas do Brasil, Lipsia, F.A. Brockhaus, 1858, p. 457.  ² Revista do Museu Paulista, no. S, vol. IV, São Paulo, 1950, p. 245.

Pedagogical approach

The project rested on four guiding principles: valuing Brazilian culture; introducing teachers and students critically to language models and conversational agents; personalizing characters according to school curricula and class interests; and digital literacy understood as a reflective, ethical, and creative relationship with AI. Five training modules were produced (AI, conversational agents, character selection, research and contours, agent personality), each delivered as video, text, and lesson plan. A knowledge sheet in eight axes (identity, historical and social context, intellectual and artistic production, influences, imaginary and representations, controversies, relationship to technology, profile tuning) structured students' research and the building of characters.

Content design and production

Juliana authored all the educational content of the project — writing the texts, building the lesson plans, and designing the training videos, from storytelling through animation to production. These materials were conceived as ready-to-use tools, accessible to teachers new to artificial intelligence.

Technical architecture

The platform integrates a web interface with public and private areas, differentiated profiles (administration, pedagogical coordination, teacher, student, visitor), a hybrid conversational engine combining InBot's non-generative technology with generative models, a collaborative character database, a configuration module, analytics tools, and a RAG system to feed agents with specific content.

Characters created

Eight conversational agents were developed: six by participating teachers — Paulina, Candido Portinari, Chico Mendes, Enedina Alves Marques, Maria Firmina dos Reis, and Paulo Freire — and two model characters produced by the Ai_Monguetá team for demonstration and training purposes: Machado de Assis and Miro. The set highlights trajectories often little-known in school and digital environments.

The Paulina case

An emblematic character of the project, Paulina was created in partnership with teacher Romulo Paulino and his students at Escola Amorim Lima. Conceived as a public-school teacher inspired by Freirean principles and articulated with the school's anti-racist and community practices, she was unveiled to the public during the Festa Kizomba in November 2025 — the school community being able to interact with her by text and voice.

Montreal applications

The project had two Montreal-based applications, allowing the approach to be transposed into a Quebec context:

Germaine Guèvremont character — creation of a conversational agent inspired by the Quebec writer, extending the Brazilian work on literary and cultural figures.

Virtual homework assistant — development of a character designed as a study companion for young people, conceived not to deliver answers but to spark reflection: asking questions, prompting hypotheses, encouraging students to put their reasoning into words.

Impact and outlook

Beyond producing a platform and agents, the project constituted a true testing ground for pedagogical experimentation, weaving together technological development, teacher training, and cultural valorization. It showed that integrating AI, culture, and education depends not only on a good tool, but also on pedagogical time, attentive listening, teacher mediation, and institutional openness to experimentation. The challenges identified — infrastructure, evaluative pressures, heterogeneity of digital literacies, scarcity of organized sources on certain figures — feed perspectives for continuity: deepening teacher training, closer support during character design, expanding the character repertoire, and strengthening public spaces where school experiences can circulate.

Resources and outreach