EdTech · Math Education
Lesson-aware AI for every student. Live in 4 days.
Embedding AI into a live math tutoring platform — from API to student experience.
How Acceleo Matemática built a role-aware AI Teaching Assistant, usage metering, and real-time analytics — all on one integrated API platform.
Acceleo Matemática
Math Tutor
Live
Lesson Planner
Live
Student Coach
Draft
<4 days
API key to live
0
Infra managed
0
Lesson stages
Full
Session history
The problem
Acceleo runs live math tutoring across Brazil. Teachers needed an AI assistant that understood lesson context and adapted per student — without building model infrastructure.
Generic chatbots lacked curriculum grounding, role awareness, and session continuity. Production-grade memory, retrieval, and safety were required from day one.
What we built
We embedded a role-aware AI Teaching Assistant via the SundayPyjamas Apps API — different behaviour for teachers vs students, lesson-stage awareness, and curriculum retrieval.
Per-student memory, graceful fallback, and a real-time analytics dashboard gave visibility without operating model infrastructure.
What we built for Acceleo
- Role-aware AI (teacher + student modes) across 5 lesson stages
- 5-surface deployment: live session, post-session, TA workspace, lesson prep
- Socratic method engine and async content generation
- Safety enforcement + analytics dashboard
Technical integration
Lesson-grounded chat, per-student memory, curriculum retrieval, and resilient fallback.
Lesson-grounded chat
POST .../chat/messagelessonContext → grounded responses
Per-student memory
POST .../memory/save-messagesthread → session continuity
Curriculum retrieval
Vector index over lessonsquery → cited answers
Graceful fallback
Local fallback routeresilient classroom experience
Outcomes
Live student sessions in under four days. Teachers run AI-assisted lessons across five surfaces without managing models or vector stores.
Students receive curriculum-grounded answers with continuity across sessions.
What this enables for EdTech
Role-aware assistants, curriculum RAG, and per-learner memory apply across learning products:
Tutoring & coaching
Stage-aware assistants that adapt tone and depth by learner role.
Corporate L&D
Grounded answers over internal playbooks with usage metering.
Certification prep
Socratic practice with memory of weak areas.
Embedding AI into your product?
Production proofs on your data — live in days, not months.