About · Technical
How parent feedback fits the stack
Short notes you add after uploads travel through research-informed frameworks and multi-agent review. The interactive flow below is optional: use it when you want to see one way data can move through the system.
Optional walkthrough: one way feedback can move through analysis.
Upload & Feedback
Parent uploads media files
- Upload drawings, videos, audio, and any other files
- 15-30 second contextual feedback via modal
- Capture emotions, effort, and time spent
- 95% parent satisfaction rate
Scientific Framework
Apply psychological theories
- Gardner: Multiple Intelligence mapping
- Dweck: Growth Mindset indicators
- Ericsson: Practice type classification
- PERMA: Well-being assessment
Multi-Agent AI System
25+ specialized AI agents for collaborative analysis
- 25+ specialized agents (OpenAI, Anthropic, Google Gemini, X.AI, Mistral, Qwen, Deepseek, Meta, Groq, etc.)
- Different models from different providers have different perspectives and training biases
- Consensus among 35+ agents eliminates individual model biases and increases accuracy
- 5 specialized domain experts (Cognitive, Creative, etc.)
- 2 statistical aggregators for data consolidation
- 1 meta-synthesis agent for final review and validation
Talent Tree Vector
Vectorize into talent taxonomy
- Vector embeddings of all development data
- Scientific talent tree structure
- Longitudinal development tracking
- Dynamic inter-ability connections
AI Curator RAG
Personalized consultation system
- RAG with semantic search capabilities
- Vector embeddings of child's history
- Personalized development recommendations
- Contextual dialogue system
Scientific Agents
High Advanced agents and education systems
- Specialized predictive AI agents
- ADHD agent system
- AI Education Materials Generator (5 subjects, 7 languages, grades 1-12)
- Cross-child Adaptive Learning
- School-connected
Outcomes we optimize for in product and research.
+40% Accuracy
Enhanced precision through multi-agent consensus and bias reduction
95% Satisfaction
Parent approval rating
85% Engagement
Active user participation
Network Effects
Collective intelligence growth
Notes you add can be mapped to established psychological lenses used in our research layer.
Gardner
Multiple intelligences
Eight intelligence types
Dweck
Growth mindset
Effort and persistence
Ericsson
Deliberate practice
Skill development
Seligman
PERMA well-being
Positive psychology
Signals in the product show how strongly we believe each insight and where models agree or diverge.
How your feedback is interpreted
Feedback processing
Short parent notes are combined with uploads and routed through multiple model passes, each with a distinct role.
Confidence scoring
Each surfaced strength gets a confidence band from agreement between passes and the strength of evidence in the artifact.
Transparent results
In the app you can see which kinds of passes contributed and at what consensus level, without raw model hype.
Build trust
See why an insight appeared and what evidence it leans on.
Track impact
Understand how follow-up uploads change confidence over time.
Data quality
Spot gaps where another short note would help the most.
Actionable insights
Parent Value Score summarizes how useful and reliable a read is for your next step.