Educational AI will soon maintain perfect memory of every concept a student has encountered, every struggle they've overcome, and every breakthrough they've achieved. This persistent knowledge layer changes how we think about personalized learning entirely.
I built Evenfield because my three children needed something that didn't exist: an AI tutor that actually remembers. Not just within a single session, but across months and years of learning. Every interaction, every misconception addressed, every strength discovered gets preserved in a memory layer that never forgets.
The technical foundation makes this possible. Evenfield runs on Next.js with Supabase handling data persistence and Railway managing deployment. The AI tutor uses Anthropic's Claude API, but the real innovation sits in the memory architecture. Each tutoring session writes detailed learning data to H.U.N.I.E., my persistent AI memory system. The tutor can reference conversations from three months ago, building on previous explanations rather than starting fresh each time.
This matters more than you might expect. Traditional tutoring, even AI-powered versions, treats each session as isolated. A student struggles with fractions in October, masters them by December, then encounters ratios in March. Without memory, the AI can't connect these concepts or reference the specific strategies that worked before. With persistent memory, it builds a comprehensive model of how this particular child learns mathematics.
The platform covers fifteen subjects: math, coding, financial literacy, entrepreneurship, AI literacy, Spanish, reading, science, and others. Each subject maintains its own knowledge thread while contributing to an overall learning profile. When my daughter shows strong spatial reasoning in geometry, the system notes this strength and applies visual approaches in her coding lessons.
Age and skill level differentiation happens automatically. The AI adjusts vocabulary, complexity, and examples based on accumulated knowledge about each learner. My youngest gets fraction explanations with pizza slices. My oldest gets algebraic representations. Same concept, different cognitive entry points, all informed by their learning history.
Compliance requirements drove another feature set. Homeschool regulations require documented progress tracking. Evenfield generates quarterly PDF reports automatically, pulling from the complete learning record. No manual compilation or guesswork about what happened three months ago. The system knows exactly which concepts were covered, how well they were mastered, and where gaps remain.
The real test isn't technical specifications but daily usage. My children actually use this system for their education. Not as a supplement or experiment, but as their primary learning platform. When your own kids depend on something working correctly every day, you build it differently. Edge cases become critical bugs. Performance matters. The interface needs to handle tired eight-year-olds and focused teenagers equally well.
Evenfield represents the first production implementation of H.U.N.I.E., my persistent AI memory system. Every tutoring session demonstrates that AI can maintain meaningful long-term memory about learners. This isn't theoretical anymore. The system tracks three children's educational progress across multiple subjects, maintaining context that spans months of learning.
The implications extend beyond homeschooling. Any educational context benefits from persistent memory: traditional schools, corporate training, professional development. The technology exists now to build AI systems that remember everything a learner has encountered. The question becomes how to structure that memory for maximum educational impact.
I built this for my children because existing solutions couldn't remember what they learned yesterday. Now they have an AI tutor with perfect recall of their entire educational journey. That changes what's possible in personalized education.
The code runs in production. The children learn daily. The AI remembers everything.
Top comments (0)