It automates the repetitive parts of teaching a language:

- Creating lesson plans. - Creating and correcting homework and language learning activities (listening, writing, multiple choice, and soon adding speaking exercises!). - Teachers get personalised feedback of the weaknesses and strengths of their students.

The goal wasn’t to “replace” teachers but to shorten the feedback loop so learning happens faster. In our pilot, feedback that used to take days now reaches students in minutes.

It’s live for ESL right now.

Sharing because I’m curious how others have handled:

1) Privacy-first design when using GenAI in public-sector education.

2) Evaluating language accuracy with LLMs without over-fitting on prompt engineering.

3) Deploying AI tools under real institutional constraints (e.g., procurement, ethics guidelines).

Happy to answer questions or share architecture details!!