Looking for feedback from the community – testing an anti-scam AI prototype

General Discussion
1

Hey everyone,

I’m working on a project called Bastian, a conversational AI designed to help people recognize and avoid scams, especially during those emotionally intense moments when panic can cloud good judgment.

This project comes from a personal place. I’ve seen people close to me get scammed, and I’ve also experienced manipulation and deception in my own life that left a deep impact. I’ve spent years studying how people get tricked, what actually happens in those moments, and I realized what we need most is something that slows us down and helps us think clearly when it matters most.

I’m not an engineer, but I pushed myself to build a working prototype of Bastian anyway. It’s designed to act like a calm, informed second voice that helps users pause, analyze the situation, and spot red flags—whether it’s a phishing message, fake support call, or something more subtle.

Right now, I’m looking for folks with real experience in this space, scambaiters, fraud victims, or people who study scam tactics, to help me test it and tell me what’s missing, what breaks, or what just feels off.

No data is saved, and I’m not trying to sell anything. I just want your insights to make this tool actually useful.

If you’re open to trying the prototype or want to know more, I’d love to share a link. Feel free to comment or DM.
And if anyone’s open to a quick call, I’d love to learn directly from the people who know this world better than anyone.

Thanks in advance

2 Likes
2

How does it work?

Did you code it, is it cloud based, run on a phone, a laptop?

1 Like
3

Right now, it’s embedded on my landing page, so it’s cloud-based and accessible through a browser. I haven’t done much deep prompt engineering yet, because I don’t have a solid baseline of what works and what doesn’t in the wild.

That said, I’ve tested it against real scam examples pulled from Reddit threads where people were asking for help, and so far it’s nailed every one. Still, I know I have blind spots, and that’s exactly why I’m hoping to work with scam baiters, to refine and test it against more edge cases. I’m not trying to add risk I’m trying to reduce it.