AI
Where a verified, tool-using AI genuinely helps — and where a simpler, non-AI approach is the better engineering choice.
Learn More →Squirrel Wisdom Research
Studying the technology underneath learning — and keeping it reachable.
A research group inside Squirrel Wisdom Academy, building the graphics, AI and connectivity technology behind modern education — real, working systems, not just product marketing.
- Three original technologies, built in-house
- Built on real, working systems
- Research that reaches real classrooms
What we study
Five research areas, one underlying question.
How do the technologies behind modern media, connectivity and AI actually change what a student can learn, and who can reach it?
Graphics & Synthetic Media
From data hidden inside ordinary video compression to real-time synthetic faces built without a generative model.
Learn More →AI in Education
What verification, tool-scoping and cost-aware design actually mean once an AI assistant is in front of a real student.
Learn More →VR & AR in Education
What changes when a lesson happens in a real place, virtually — presence, engagement and shared attention at a distance.
Learn More →Accessibility & Reach
Keeping educational content and communication reachable when the network — or the hardware — won't cooperate.
Learn More →
What we've built
Three breakthroughs, one research program.
A High-Throughput Video Modem — we built a way to carry an independent data channel inside an ordinary video call, invisible to the compression that normally destroys anything hidden in the picture.
A Data Channel Built Into Compression Itself — a blind data channel engineered directly into the mathematics of H.264 video compression, built to survive it rather than merely tolerate it. Developed with support from the Open Technology Fund.
Facial Mimicry Cloning — real-time synthetic video built with no generative model at all: real recorded frames, intelligently recombined. See Graphics & Synthetic Media for the full story.
How we work
A few principles that shape everything here.
Real, working systems
Every idea here gets built into working software and tested end-to-end — not left as a slide deck or a research memo.
AI where it earns its place
We study where a neural network genuinely helps, and we're just as interested in the cases where a classical, non-AI approach wins on cost, latency or honesty.
Built to run on ordinary hardware
No result here depends on a GPU cluster a typical school could never afford — that constraint is treated as part of the research question, not an afterthought.
Research that reaches the classroom
Every line of work here connects back to a real product a student or teacher can actually use — see how in AI in Education and VR/AR in Education.
Want to follow or collaborate on this research?
We're glad to talk to schools and fellow researchers alike.