VishAlyze.ai

Inspiration

I was inspired to create this project because of my years of industry experience in cybersecurity. I’ve seen firsthand how traditional phishing simulations—especially for voice phishing (vishing)—fall short by failing to test how employees actually respond when pressured. I wanted to build a smarter solution that captures not just awareness, but real-time behavior and decision-making under threat.

What it does

VishAlyze.ai is an AI-powered vishing simulation and training platform. It:

Delivers realistic, AI-generated scam calls to employees

Captures and records how they respond

Analyzes verbal responses using GPT and speech-to-text

Provides personalized feedback and risk scoring

Gives security teams a full dashboard to manage campaigns, listen to recordings, and build training content from real scenarios

How we built it

Voice Generation: We use ElevenLabs to create natural-sounding, AI-driven scam call audio

Call Delivery & Recording: Twilio powers programmable calls and voice recording

Speech Analysis: We use Whisper (OpenAI) to transcribe user responses

Feedback Engine: GPT-4o analyzes tone, word choice, and security posture

Dashboard: Built with React + Supabase for managing targets, reviewing calls, and launching campaigns

Challenges we ran into

Designing realistic but ethical scam call scenarios

Tuning AI feedback to be actionable without false positives

Handling variability in human speech and noise during transcription

Keeping the UX simple for cybersecurity teams while offering deep insights

Accomplishments that we're proud of

Creating the first simulation that actively records and learns from verbal responses

Building a complete voice threat simulation pipeline end-to-end in a short time

Designing a gamified feedback loop to make training engaging and effective

Crafting a tool that’s easy for security teams to deploy but powerful enough for real-world resilience

What we learned

Response behavior is just as important as detection in social engineering defense

Voice phishing is still an underserved but growing threat vector

With the right AI tools, we can go beyond training and into real behavior modeling

Security awareness platforms can be smart, engaging, and measurable

What's next for VishAlyze.ai

🎧 Real-time coaching mode: Let users get feedback live during simulated calls

🎮 Gamified team competitions: Engage entire organizations in weekly phishing challenges

🤝 Partnerships with MSPs and security vendors

📈 Enterprise reporting & audit logs for SOC2, HIPAA, and ISO compliance

Built With

• bolt
• chatgpt