Inspiration
Scams are getting more and more sophisticated and it's not enough to just protect people. We need to fight the roots of this evil. AI keeps getting more and more realistic, cheaper and easier to use. Therefore if we don't do anything about it, scams will keep getting more and more widespread.
What it does
Our software serves as a layer of active protection between scammer and non-tech-savy people. It will actively take calls, interrupt emails and text messages for the sake of wasting scammers' time. With such software integrated into every phone and email provider we will make scamming impossible. We will waste so much of the scammers time that the scamming itself will stop making money and will seize to exist. We will make sure than scammers's time is wasted and they are reported to the authorities.
How we built it
Wired Gmail API OAuth (credentials.json + token.json) to fetch unread messages. Sent message content to an n8n workflow that classifies senders via GPT. Persisted sender status (whitelist/blacklist/none) in Supabase PostgreSQL. Automated replies for blacklisted senders with time-wasting, humorous templates. Added optional Redis for lightweight caching and throughput.
Accomplishments that we're proud of
End-to-end pipeline from Gmail → n8n/GPT → Supabase → auto-replies. Robust scam engagement that convincingly wastes scammers’ time without risking users. Clear sender lifecycle: unknown → blacklist/whitelist with learning over time.
What we learned
Strong, explicit system prompts improve consistency in classification and replies. Conservative defaults and reproducible workflows reduce noisy or ambiguous outcomes. Simple states (whitelist/blacklist/none) make automation auditable and robust. Humor and confusion patterns are effective for long scammer engagement.
Built With
- gmailapi
- javascript
- n8n
- openaisdk
- postgresql
- python
- redis
- supabase
- twilio
Log in or sign up for Devpost to join the conversation.