SIGNAL — AI Trust Intelligence Platform

AI-powered platform that detects fake jobs, internship scams, phishing emails and misinformation in real time using forensic analysis, back-search, and psychological vector breakdown.

Comment

The Problem That Started It All

I once received a fake interview call myself — it felt completely real until it didn't. That experience stayed with me. Every day peoples , students and freshers lose money, time, and confidence to fake job offers, internship scams, phishing attacks and even the fake news or misleading information which tricks peoples to believe in fake messages . Existing tools give a simple yes or no verdict. I wanted something that explains why something is dangerous — like a forensic investigator working in real time.

What SIGNAL Does

SIGNAL is an AI-powered Trust Intelligence Platform that analyzes text, URLs, emails, image posters, and files to detect scams, phishing, and misinformation in real time.

  • Delivers a forensic trust score displayed as a precision radar — showing exactly how safe or how risky the content is, not just a binary verdict
  • Shows a clear final verdict — a single decisive message telling you whether to proceed or not proceed, with the confidence level behind it
  • After every scan, a contextual chatbot is available that already knows everything about the scanned message — users can ask follow-up questions to understand exactly why something is safe or unsafe, going deeper than what's shown on screen
  • Built-in Quick Demo scenarios in the nav — users can instantly load example messages for verified jobs, scam offers, spoofed domains, fake news, and more to see SIGNAL in action before scanning their own content
  • Supports 10 globally dominant languages — English, Hindi, Chinese, Spanish, Japanese, German, French, Portuguese, Arabic, and Korean — with full UI translation and forensic reports generated in the user's chosen language, making SIGNAL accessible worldwide
  • Comparison mode — evaluate up to 5 job offers or internships side by side, with individual forensic results and a final best-choice recommendation

How I Built It With MeDo

I used MeDo's Deep Build feature across 39+ iterations — zero code written, entirely through prompting. I structured multi-turn conversations to progressively build each layer: the detection engine, forensic output format, comparison mode, multilingual support, session history, and fully responsive UI across all devices.

Most Impressive Feature

The forensic engine classifies content type, runs live back-search across the web, scores each red flag individually with precision percentages, identifies psychological manipulation vectors — urgency, financial lures, authority spoofing, cognitive dissonance — cites real external sources like Lifehacker, Indeed, and LifeLock, and delivers a complete forensic summary from a single paste.

Important: SIGNAL deliberately prioritizes accuracy over speed. Most scans complete within 45 seconds. This is intentional — a deeper, more thorough analysis is worth the wait when the stakes are protecting someone from a scam.

Challenges

  • Managing edge function timeouts caused by the depth of multi-step analysis — solved by breaking the scan into sequential stages with retry logic
  • The app crashed multiple times during development due to the sheer volume of features being built simultaneously — required carefully structured prompts to isolate and fix each issue without breaking existing functionality
  • Communicating complex, layered requirements to MeDo with enough precision that it understood exactly what to build — learning to prompt like an architect, not just a user
  • Making all 10 languages work seamlessly across both the UI and AI-generated forensic reports
  • Getting the URL scanner to handle fetch failures gracefully with retries and clear user feedback

What I Learned

Prompting an AI builder with precision is a skill in itself. I learned to think in systems — breaking features into layers, describing behavior not just appearance, and iterating with purpose. The biggest lesson: you don't need to write code to build something serious. You need clarity of vision and the patience to refine it.

What's Next for SIGNAL

  • Browser extension so users can scan content without leaving any webpage
  • API access for platforms to integrate SIGNAL's trust layer directly
  • Expanding to scan voice messages and video job interviews for deepfake detection
  • Building a live scam database crowdsourced from SIGNAL users worldwide
  • Partnerships with job platforms to flag suspicious listings before they reach candidates

Built With

Share this project:

Updates

Submission history