ReliefAI
Here the prompt work is a safety problem, not a quality problem.
Try it live- Safety under distress
- Returning loop
Every other product on this page fails by being unhelpful. This one can fail by saying the wrong thing to someone in real distress, so tone and safety are the product, not a coat of polish on top of it.
Support for people going through a breakup: empathetic conversation, daily coping exercises, reflective journaling and a set of assessment tools. This needed the deepest prompt architecture of anything I have shipped, because the failure mode is not a bad answer, it is a harmful one.
What I built
- Prompt architecture tuned for tone, safety and emotional accuracy under real distress.
- Assessment tools that turn a vague feeling into a clearer next step.
- Daily exercises and journaling as a returning loop, not a one-shot chat.
Inside the product
