Locker
An offline-first AI study companion that high-schoolers install once and use for years — homework help, reading summaries, and practice problems across STEM subjects without sending any data to the cloud.
High-school students preparing for exams in homes with patchy or monitored internet
- Installable desktop app that runs an 8B model locally after a one-time download
- Subject packs (algebra, biology, world history) with curated practice problems and worked solutions
- Reading-level adapter that re-explains any passage at a chosen grade level
- Parent-side dashboard showing time-on-task topics without exposing conversation content
The Neutrino-1 release shows compact open models can run on a normal laptop, opening the door for a real offline student experience just as districts debate banning AI tools that send student data to third parties.
Demand is real and validated by KernelAI explicitly targeting the exact same use case (offline homework help, exam prep, STEM, privacy) on iPhone, plus multiple 2025 roundups of AI homework helpers; FERPA debates and Texas district bans confirm the privacy-first angle resonates.Offline AI for Students - Free AI Tutor That Works Anywhere | KernelAI ↗FERPA in the Age of ChatGPT: What Every School District's IT Leader Should Know ↗
KernelAI is a near-exact competitor: offline, on-device, free, multi-subject, privacy-first AI tutor for students with 44 bundled open-source models; open-source alternatives (study-buddy, AI-Tutor on GitHub) further crowd the space.Offline AI for Students - Free AI Tutor That Works Anywhere | KernelAI ↗GitHub - michael-borck/study-buddy: Provides an offline AI tutoring ↗GitHub - hari7261/AI-Tutor - Local AI Study Buddy ↗
KernelAI ships the same pitch for $0 forever and explicitly compares itself to ChatGPT Plus at $20/mo, anchoring consumer willingness-to-pay near zero; viable monetization likely requires B2B district licensing or premium subject packs rather than direct consumer subscription.Offline AI for Students - Free AI Tutor That Works Anywhere | KernelAI ↗
Privacy/regulation tailwind is durable (FERPA, COPPA, state student-privacy laws, ongoing district debates and outright bans) and local-model capability keeps improving per Fermion's Neutrino-1 8B ternary container claiming laptop/CPU viability.Fermion Neutrino-1 8B — 3.88 GB Ternary, 72.1 MMLU ↗FERPA and AI: Can Schools and EdTech Use ChatGPT With Student Data? ↗
Technically proven feasible by KernelAI running 3B models on iPhone offline with 44 bundled models; cross-platform laptop support, bundled model updates, subject-aligned curriculum, and app-store distribution add non-trivial complexity.Offline AI for Students - Free AI Tutor That Works Anywhere | KernelAI ↗