TeachMap AI Tutor Now Remembers Your Learning Journey

TeachMap AI Tutor now has a real sense of continuity. Students can return to past tutoring sessions, rename conversations, delete conversations they no longer need, and keep working from the same learning thread instead of starting from a blank chat every time. This update also brings the same material-aware workflow to both sides of the tutor. The text tutor and voice tutor can work with uploaded study materials, including documents, spreadsheets, slides, and images. A student can ask about a worksheet in text, come back later, restore the session, and continue with voice when talking through the problem is easier. The biggest change is quieter: TeachMap AI can now save small, durable learning memories. These are not full transcripts. They are concise learning signals that help the tutor adapt across future sessions.

• Session history for Tutor AI conversations
• Rename and delete controls for saved tutoring sessions
• Document and image uploads available to text and voice tutoring
• Restored sessions carry recent context forward for follow-up questions
• Small learner memories help future sessions feel more personalized

Most real learning does not happen in one sitting. A student might ask for help with linear equations on Monday, revisit the same homework on Tuesday, and need one more explanation before a quiz on Friday. Without history, every tutoring session has to rebuild the context from scratch. Tutor history changes that. Past sessions now remain available in the tutor, so students can reopen the conversation that belongs to the assignment, topic, or lesson plan they were studying. If the title is not clear enough, they can rename it. If the session is no longer useful, they can delete it. For voice tutoring, the restored conversation still matters. Voice exchanges are saved as text turns behind the scenes, which means the tutor can use the prior back-and-forth as recent context even when the student switches from typing to speaking.

History and memory are different. History saves the conversation so a student can return to it. Memory distills what may matter later. TeachMap AI looks for durable learning signals after a tutor exchange finishes. It may remember that a student prefers visual examples, is working toward a specific goal, has mastered a concept, or still has a misconception that should be handled gently next time. These memories are short, usually one sentence, and they are saved only when they are likely to help future tutoring. The system is intentionally selective. It does not need to remember every question, every answer, or every casual detail. The goal is for the tutor to feel like it is learning how to teach the student better, without turning the database into a pile of raw chat logs.

• Memories are distilled after the answer is sent, so tutoring stays responsive
• Each exchange can produce zero, one, or a few concise learning signals
• Private personal details, payment data, and one-off chatter are filtered out
• Similar memories are merged or updated instead of duplicated forever
• Only the most relevant memories are retrieved for each future question

Students rarely learn from a clean prompt alone. They learn from worksheets, textbook pages, teacher slides, handwritten notes, lab tables, essay drafts, and screenshots. Tutor uploads are now part of the shared Tutor AI workflow instead of being limited to one mode. That means a student can upload the actual thing they are studying and ask for help grounded in that material. For text tutoring, this is useful when they want a written explanation they can scan. For voice tutoring, it is useful when they want to talk through the same material step by step. This is especially helpful for families and classrooms where the adult nearby may not know the subject. The student does not have to retype the whole problem or summarize a diagram from memory. They can give the tutor the material and ask the question directly.

A good tutor does more than answer the current question. It notices the pattern behind the question. It remembers that a student keeps mixing up slope and y-intercept, or that they understand the big idea but need help organizing written answers. It adjusts the next explanation because the last one taught it something. That is the direction TeachMap AI Tutor is moving. The tutor can now carry enough context forward to make future sessions feel less repetitive and more personal. A student should not have to say, "I always need examples," every time they come back. They should feel the tutor gradually adapting. For students with seven classes a day, the design has to stay compact. A few useful memories per class are enough. The tutor needs signals, not clutter.

Tutor memory is backed by TeachMap's database and vector search, so future questions can retrieve the most relevant learner signals instead of loading every saved note. This keeps the tutor focused on what matters for the current question. The memory system also has sensible limits. Active memories are capped, similar memories are merged, and older lower-priority memories can be archived as the learner profile evolves. That gives students room to grow across subjects without letting old or duplicate notes crowd out better ones. The result is a tutor that can remember enough to feel continuous, but not so much that it becomes noisy. It is designed for real students, real school schedules, and the practical messiness of learning over time.

• Vector retrieval finds relevant memories for the current question
• Memory writes happen after the tutor responds, keeping chat and voice flows fast
• Active memory is bounded so the learner profile stays useful
• Old or duplicate memories can be replaced as the student changes
• The same approach supports both general tutoring and lesson-plan-aware tutoring