Add your textbooks, notebooks and syllabus to AI-Memory once. From then on Grout writes lesson plans, notes, slides and worksheets from the book your class actually uses — not from whatever a model happened to absorb during training. Everything runs on your own machine.
AI-Memory organises sources the way a syllabus is organised: a subject holds textbooks, notebooks and reference material, a textbook splits into chapters, a chapter holds its pages.
Add the PDFs you already have, or photograph the pages of a physical textbook and let the OCR read them. Split the book into chapters so a request can be pointed at chapter four rather than at the whole volume. Add your own handwritten notebooks alongside it, and the board or university syllabus as a separate source. Indexing happens on the machine — nothing is uploaded, so a prescribed textbook never leaves the building.
Point Grout at a chapter and ask for what you need. Because the answer is grounded in your source, the terminology and the worked examples match the book in front of your class.
Ask for a lesson plan on a chapter and get one built from that chapter's definitions and examples rather than a generic outline. Pull the same material into class notes at the length you want, then lay it out on the canvas editor with generated diagrams and illustrations where a picture explains it faster. Export the result as a PDF, an image set or a GRT project to reuse next year.
Generate practice from the material you have already loaded, at more than one level, so the same chapter reaches the whole room.
Ask for a question set on a chapter and specify the level. Generate a second version for students who need more scaffolding and a third for those who are ahead, all drawn from the same source so nobody is working from different definitions. Because the model reads your book, the notation matches what students have already seen.
Students write on paper as they always have. Grout comes in at the marking stage: scan or photograph the scripts and let the on-device vision models read them.
Scan a batch of handwritten scripts and mark them against your own answer key and rubric. The vision models read handwriting and subject notation, draft feedback per answer for you to edit before it goes back, and surface the questions a whole class got wrong so the next lesson can address the gap. Scripts are processed locally — no student's handwriting is uploaded anywhere.
Start of term
Load the prescribed textbook, your own notebooks and any reference material into a subject in AI-Memory, and split each book into chapters.
Once per subject
Add your board or university syllabus as a source so units and course codes line up with the chapters they are taught from. Grout ships curriculum packs for KTU, VTU, Calicut, CHRIST, CBSE and CISCE if yours is already covered.
Every week
For each week, ask for the lesson plan, the notes, the slides and the worksheet from the chapter you are teaching. The setup is done; the rest of the term is requests against material that is already indexed.
Grout runs on your own desktop, offline, with your textbooks and notebooks indexed locally. One licence covers all three apps.