A lot of us know someone who's gone through a hard time and didn't know who to talk to. Crisis lines are great, but sometimes there's a wait, and honestly calling a stranger is scary. We wanted to make something you could just open and start typing to, any time, so we built Signal. None of us had really worked with LLMs or APIs before this, sot here was a lot to learn. We figured out how to send requests to an LLM API, how much the system prompt changes the way the model responds, and why you should never put your API key in your frontend code. Signal has a simple chat interface that sends messages to our backend, which calls the LLM and sends the response back. We spent a lot of time on the system prompt so the bot sounds supportive and doesn't try to act like a therapist. Getting the tone right was way harder than we expected, since at first the responses were either super long or felt robotic. We also kept running into API errors and had to figure out how to handle them so the app didn't just crash. In the end, we're proud that we built something that could actually help someone.