01 /
Teaching Intelligence
Instructional examples showing how experts sequence explanations, select scientific visuals, and assess understanding of complex concepts.
Scientific Data Intelligence
Intelligencebegins withevidence.
Building the data foundation to teach AI how to teach science—by preserving how experts source, explain, visualize, assess, and revise scientific concepts.
The Opportunity
The most valuable data isn't just stored. It's connected.
A finished scientific video is only the surface. Behind it lies an entire system of scientific teaching:
- Scientific sources & references
- Expert storyboards & instructional sequences
- Medical illustrations & narration
- Educational videos & assessments
- Historical revisions & source versions
Together, these connections reveal how experts turn scientific knowledge into understanding.
What we know. How we teach it. How it evolves.
The Foundation
A universe of scientific material.
A deeply layered archive of scientific sources, instructional design, multimedia education, and historical revisions—now being mapped into connected data for AI.
- 01Inventoried objects
- 1,516,140
- 02Source repository
- 2.63 TB
- 03Identified modules
- 569
- 04Current screens identified
- 11,628
Preliminary repository census. Dataset qualification and rights assessment are ongoing.
Our Approach
Structure changes everything.
STAGE 01 / 03
01
Discover
Identify the current published instructional content, its underlying scientific sources, and the historical versions that document its evolution.
02
Connect
Reconstruct how experts transform scientific evidence into storyboards, sequenced explanations, medical visuals, narration, videos, and assessments.
03
Enable
Build version-aware datasets that can help AI learn how to teach scientific concepts—and adapt instruction as its source material evolves.
What This Makes Possible
Built for the next frontier of scientific AI.
Strategic directions under exploration — not production products.
02 /
Model Evaluation
Potential benchmarks for scientific grounding, instructional clarity, multimodal alignment, and responses to changing evidence.
03 /
Scientific Knowledge Infrastructure
Reconstructing published sources of truth, their provenance, instructional relationships, and historical revisions.
04 /
Data Partnerships
Exploring AI training, evaluation, licensing, and research collaborations built around qualified scientific learning data.
Dataisthefoundation.Teachingistheintelligence.
About Onavi
A new perspective on scientific intelligence.
Scientific learning repositories contain more than knowledge. They preserve traces of how experts turn that knowledge into understanding—and how instruction changes over time. Onavi is working to connect scientific sources, teaching structures, multimedia, and historical revisions into a foundation for AI that can learn not just science, but how to teach it.
Let's Connect
The next breakthrough starts with better data.
We're exploring collaborations with AI researchers, model developers, and scientific organizations interested in teaching intelligence, multimodal learning data, and version-aware scientific knowledge.