ScamWhiz

Inspiration

Each year, millions of people fall victim to online scams and consumers loose billions of dollars to scams and frauds. Blackmail scams can cause victims to feel humiliated, scared, and isolated.

Scammers often use similar tactics and repeatedly target multiple people with their schemes thereby creating a pattern that can be spotted. With a simple search in the scam database using generative AI tools, before taking any further action, a red flag and scam can be spotted using ScamWhiz.

What it does

This is a GPT-4-Mini-powered scam, radar, finder and reporting app with an integrated generative AI chatbot.

Report a scam and find a scammer online. Enter a text of any suspicious Message, Username, Pseudo Name, Email, Phone Number, crypto address, website, or DM, and ScamWhiz will let you know if it's a scam or not.

How we built it

Response streaming was implemented via the Vercel AI SDK

Clerk is used for authenticating users.

MongoDB was used for storing scam reports.

Pinecone was used for storing the vector embeddings of the scam reports

Open AI SDKs/APIs, GPT-4o-mini and text-embedding-small models were used for the chats and embeddings.

Accomplishments that we're proud of

Bringing this application to life.

What we learned

I learnt the Open AI SDK, databases and other frameworks used.

What's next for ScamWhiz

• Search by picture or screenshot, audio or video
• Support more file formats
• Pull in documents via API
• Add search with AI on in the input field
• Add search with AI off in the input field
• A mobile version of the app on android

Built With

• gpt
• mongodb
• next
• openai
• pinecone
• text-embedding-small
• typescript
• vercel