Our project was inspired by Dario Amodei’s essay "Machines of Loving Grace," which imagines AI as a transformative force for health and mental wellbeing. We identified a "clinical dead-end" in current AI: while tools like OpenAI and Gemini offer information, they lack human warmth and often isolate users in a digital bubble. We chose the name Sankofa—an Akan concept meaning "to go back and fetch"—to represent reclaiming one's identity and resilience from the wreckage of a crisis through human connection.
How We Built It We focused on Track 2: Neuroscience & Mental Health to expand access to peer support and education
The Brain: We used the Claude API to act as a "Listener Agent" that triages user struggles.
The Platform: We chose WhatsApp for its high accessibility, ensuring our solution is "for all, not just for techies".
The Network: Instead of a chatbot giving advice, the system queries a database of human survivors and matches them with the user.
Safety First: We implemented a "Red Line" trigger that detects high-risk trauma (like abuse or self-harm) and bypasses the peer network to provide immediate professional hotlines.
Challenges We Faced
The 4-Hour Sprint: Building a working feature in such a short window forced us to prioritize a "proof of concept" over a finished product.
Ethical Guardrails: Designing a system that knows exactly when to hand off to a real professional was a critical technical and moral challenge.
Privacy vs. Connection: Balancing the need for anonymous triage with the goal of direct human contact required careful logic design.
What We Learned We learned that the most powerful use of AI isn't to replace humans, but to act as a bridge to them. We discovered that by using AI to facilitate peer support, we can tackle the therapist shortage and the stigma surrounding mental health on campus. Ultimately, we proved that AI can be a "force for good" by helping people fetch back the parts of themselves they thought they had lost.
Log in or sign up for Devpost to join the conversation.