The problem Chain of thought solves

In many Asian cultures, mental health carries a heavy social stigma. Seeking professional help can lead to gossip, loss of trust, and even discrimination from family or employers. Even if someone is willing to seek help, a deeper issue remains. They cannot trust the system to keep their struggles truly private.
Clinic records, online consultations, and even casual conversations in healthcare circles can lead to unintentional leaks. Chain of Thought solves this by giving people a safe, private, and stigma-free way to seek mental health support.
By combining Oasis Sapphire’s privacy-first blockchain with ROFL’s confidential AI execution, every interaction is encrypted, processed inside secure enclaves, and stored in user-owned smart contracts. Only the user can access their history. This creates the first mental health platform that is private by design, culturally sensitive, and accessible without fear of exposure.

Challenges we ran into

Scaffold-ETH refused to expose a default port. The lack of clear reference code in the documentation meant we had to dive deep into its source, scour community discussions, and experiment until we got a working configuration.
Our indexing layer also proved tricky. We wanted to use The Graph Studio to track privacy-safe events, but it didn’t support the Sapphire Testnet. This forced us to run The Graph locally, which came with its own set of headaches: configuring the node, syncing with Sapphire, and deploying subgraphs were all far more complex than expected. Without much official guidance, we ended up crafting our own setup scripts and Docker configurations to make the process repeatable.

These infrastructure challenges were compounded by our privacy-first design. Running AI inside a Trusted Execution Environment via ROFL required a secure way to pass encrypted user messages in and out of the enclave without ever exposing plaintext. Achieving this meant building a custom message-handling pipeline and optimizing our encryption process so users could still enjoy a smooth chat experience.

In the end, each of these hurdles pushed us to better understand our tools, improvise solutions, and document our process so the next developer walking this path won’t have to start from scratch.