Our inspiration came from one of our grandmothers, who lost money to a fake eBay refund scam that pressured her to act before she could check with anyone. We built Guardian Circle so that older adults have a plain-language second opinion inside the inbox they already use, without needing to learn a new app. Using Superhuman Go’s Agent Builder connected to Gmail, we wrote custom instructions that look for combinations of red flags, such as impersonation, urgency, refund or payment requests, and secrecy, rather than single keywords. For each message, the agent gives a High, Medium, or Low risk rating, explains the specific evidence, and recommends a safe next step, while leaving every decision to the user. Our biggest challenge was avoiding false positives, since real family emails can be urgent too, and an agent that cries wolf quickly gets ignored. We also worked hard to keep explanations short, calm, and jargon-free so they actually help instead of alarming people. We learned that pointing to exactly why a message is suspicious builds far more trust than a simple “SCAM” label, and that protecting someone’s independence matters as much as protecting their money.
Built With
- collections
- csv
- json
- pathlib
- superhumango
- sys
Log in or sign up for Devpost to join the conversation.