KinderAI Buddy: An Agentic, Multimodal AI Companion – A Research Framework for Safe and Personalized Child Learning

This paper introduces KinderAI Buddy, an agentic and multimodal AI system designed to deliver safe, personalized learning experiences for children aged 5–12. While conversational AI has gained traction in education, most existing systems lack continuity, adaptability, and ethical safeguards tailored to young learners.Our framework addresses these gaps through an agentic learning loop that incorporates self-reflection, memory extraction, adaptive goal setting, and skill-based difficulty tuning. The system integrates advanced AI technologies—including OpenAI (ChatGPT, Whisper), ElevenLabs, and MediaPipe—alongside facial recognition to enable engaging interactions across voice, video, and text modalities.We present the overall architecture, agentic flow, and ethical design considerations underpinning the framework, highlighting how KinderAI Buddy supports safe, personalized, and engaging child-centric learning. This research lays the foundation for future development of trustworthy AI companions that enhance learning continuity, safety, and personalization in education.