AI powered mental health support for youth
As youth began to increasingly turn to AI for emotional support, Kids Help Phone recognized that public AI tools often lacked the empathy and safeguards needed for mental health support. In response, we co-created an Agentic AI prototype designed to demonstrate how technology can responsibly and thoughtfully extend existing care models.
The challenge
For Kids Help Phone, the goal wasn't just to build another chatbot, but to explore how AI could meaningfully extend their model of care. The challenge was to design a digital entry point capable of engaging youth while upholding rigorous clinical standards, especially in high-risk conversations.
Our approach
We took a product-thinking approach focused on defining a production-ready path for responsible AI within a compressed timeline. In collaboration with Kids Help Phone counsellors and their digital team, we prioritized rapid prototyping to build a testable model that could clearly demonstrate the concept in action.
We built a multi-layered AI system that blends emerging technology with empathy.
Prototyped a safe and trusted entry point for mental health support
We designed a clinically guided multi-agent AI prototype to act as a safe space for youth to start a conversation. This architecture independently assesses emotional tone and risk through specialized Classification and Judge agents, ensuring that AI interpretation is guided by clear clinical boundaries rather than inference or guesswork. To ensure absolute privacy and safety during this phase, no identifiable or protected youth data was used in the development or testing of the prototype.
Created escalation pathways for human-centered care
By building a functional model with mandatory escalation logic, we proved that AI can safely augment human care. The prototype provides a seamless handoff to trained counselors, demonstrating how to reduce strain on crisis services while prioritizing human intervention where it matters most. This work is a high-fidelity prototype intended for continued exploration. As Kids Help Phone continues to test and evolve this toolkit, its specific parameters will be refined to keep pace with the changing landscape of youth support.
Moving from ambitious thinking to decisive action through rapid prototyping to prove a concept works.
Established the foundation for executive buy-in and national visibility
The speed and success of this work provided the evidence Kids Help Phone needed to secure executive support and national visibility. We defined a production-ready path that helped mitigate risk while setting a new and higher standard for responsible AI in the mental health sector. Through close collaboration, the final prototype stayed true to KHP’s values while building real-world validation for the technology.
Key Highlights
Clinical framework grounding
AI responses were trained, tested and audited against KHP’s established clinical guidelines.
Safety-critical validation
Proved the system’s ability to support youth while maintaining strict risk-aware escalation pathways.
Agentic governance
A dedicated "Judge Agent" independently evaluates and rejects unsafe responses, ensuring clear clinical accountability.
Rapid venture prototyping
Moved from strategy to a validated, bilingual, and privacy-conscious framework in a compressed timeline.
Ready to make your digital experience exceptional?