About the project
A study workspace built around measurable behavior
Lumivox is a bilingual study and productivity platform created as a final-year university project. It brings planning, focused work, reflection, and AI-assisted guidance into one connected system.
What Lumivox is
Lumivox helps learners turn goals into tasks, complete focused study sessions, and review the behavioral patterns created by that work. The product is designed for students, independent learners, and knowledge workers who need more than a static to-do list.
Its central idea is simple: productivity tools become more useful when they explain patterns instead of only counting activity. Lumivox connects task completion, deadline behavior, focus time, distractions, and reflection so users can see how their routine changes over time.
The problem
Traditional planners are good at recording intention but often stop there. A completed checkbox says what happened, but it does not explain whether a workload was realistic, when concentration was strongest, or which tasks were repeatedly delayed.
Lumivox addresses that gap with behavioral analytics and contextual recommendations. The aim is not to automate personal judgment. It is to give learners clearer evidence for deciding what to work on next and how to adjust a study routine.
Approach and audience
The platform combines goal-linked tasks, focus sessions, weekly reflections, study rooms, learning roadmaps, and document-grounded AI. A native task-risk model estimates the likelihood that a task may be late within a 14-day horizon and presents that signal as decision support.
The experience is intended for people who want structure without aggressive productivity pressure. English and Vietnamese interfaces make the same core workflow available to both language groups.
Project origin and design philosophy
Lumivox began as a final-year project exploring how behavioral data and machine learning can support self-regulated learning. The project treats implementation, model evaluation, usability testing, and transparent communication as parts of the same product problem.
The design principle is calm intelligence: recommendations should be understandable, evidence should be labeled by its real source, and the interface should help users act without overstating what the system knows.
Technology
The web application uses Next.js and React, with Supabase for authentication and relational data. AI services are implemented with FastAPI. The task-risk classifier is evaluated from a versioned artifact and registered model metadata, while generative features use configured external AI providers where needed.
This public site documents the product capabilities, research methodology, privacy model, and current limitations so the project can be assessed as a working system rather than a collection of marketing claims.