Sendero: helping Noris prepare each child's next step with local open AI
This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend . What I Built Noris teaches catechesis. She needs a record of each child's journey: when they arrive, the formation they receive, which requirements they have met, and when they move to a new stage. I built Sendero , a local companion that turns that record into preparation for the next meeting. Its Spanish…
Noris, a catechist, requires a system to track each child's journey through various stages of formation. She needs to know when they arrive, what requirements they have met, and when they advance to a new stage. To assist Noris, I developed Sendero, a local application that compiles this information into preparation for upcoming meetings.
The app is designed to be user-friendly, with a Spanish interface that integrates periods, groups, participant profiles, responsible adults, attendance, observations, stage requirements, and a chronological history. At the start of each child's profile, an open AI feature offers two or three suggested preparation actions and a question for the child's responsible adult, based on the current requirement states and attendance counts.
The proposed actions and question are saved with their model and evidence snapshot. If any record changes, Sendero flags the proposal as outdated. When creating a new period, Noris can enroll participants with multiple responsible adults, record attendance and class topics, and review the complete history. Existing profiles remain intact during schema upgrades, and new enrollments do not duplicate participants.
Noris can mark an AI proposal as reviewed, which records the review without altering the child's progress. The ultimate goal is to help Noris prepare for the next meeting, while still allowing her to make decisions about the child's progress. The prototype was tested with fictional records, and further feedback from Noris is pending.
The application is available in Spanish and features a 33-second demonstration showcasing the complete workflow. The code is open-source, using Python, Flask, SQLite, JavaScript, and the Ollama local inference model. This setup ensures that recording and inference can occur offline, without relying on external services or paid APIs.
The application follows strict validation rules to maintain data integrity and prevent invalid actions. The project demonstrates the potential of open innovation, as it can be adapted to different local environments and models while respecting each model's license.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
