Persona Lingua helps language teachers match learner personas with material from their school's library, and generates new personalized material when the library falls short.

Persona Lingua is an EU-funded project co-created by Molinos de Viento, Elite Tutors and IPHIGENIA Centre for Hellenic Studies. insidethebox is the consortium's technical partner: we designed and built the platform.

Challenge

Vast quantities of learning material is often overlooked by teachers trying find the perfect match for their student’s specific needs. The decision depends on level, on what last month's progress report said, and often on a learning barrier such as dyslexia or low literacy.

Strategy

We split the decision in two. A language model is good at reading a teacher's notes and reports and saying what a learner needs next. It is not a reliable way to rank a library. So the model writes a needs profile, a rule-based scorer ranks the school's own material against it, and the model only chooses among the top candidates and explains each choice. Material in the wrong language, already completed or too advanced never reaches it. This hybrid deterministic + AI algorithm narrows down the results and finds the best candidate.

The ranking depends on a well-described library, so uploads are read and tagged automatically by topic, CEFR level and skill, and the teacher corrects the tags before saving.

Build

Teachers keep a profile for each learner: target language, level, notes, progress reports and any learning barriers. From it the platform produces three things. Recommendations from the school's library, each with a reason. A lesson footpath: a staged plan for teaching this learner, with activities, signs of progress and things to watch for. And new material written for the learner, which the teacher can refine with further instructions and download as documents.

Library insufficiences are detected and teachers are prompted to enrich the library for better results.

The platform is language agnostic, meaning it can work on any language as long as there is sufficient amount of materials.

Outcome

Persona Lingua is used at 10 schools by 20+ teachers working with 100+ students, and has produced 200+ learning paths.

  • Schools

    10

  • Students

    100+

  • Teachers

    20+

  • Learning paths created

    200+