CISA. Hacker News. Krebs. BleepingComputer.
Same routine. Same problem.
Too much noise. Too little signal.
So I stopped reading.
And built something better.
The System
I created an AI-powered cybersecurity agent that:
- Runs every 30 minutes
- Collects real-world cyber threat data
- Summarizes it in plain English
- Classifies severity
- Sends me alerts directly to my phone
No dashboards. No scrolling.
Only what matters.
The Moment It Clicked
The first time I got a Critical alert on Telegram, it wasn't just a notification.
It was a real vulnerability. A real threat. Happening right now.
That's when I realized:
This is how cybersecurity should be learned.
Not from textbooks. From live systems.
Why I Built This
Cybersecurity is not static.
- Zero-days don't wait
- Ransomware doesn't pause
- Attackers don't follow syllabus
Yet most people learn it like theory.
This project fixes that.
Under the Hood (Simple Version)
- AI: Groq + Llama 3.1
- Backend: FastAPI
- Scheduler: Runs every 30 minutes
- Alerts: Telegram bot
- Deployment: Railway(render change soon)+ Vercel
Total cost: $0
What Changed For Me
I don't "study cybersecurity" anymore.
I observe it.
Every alert → I investigate Every CVE → I learn Every attack → I understand patterns
This system became my teacher.
What I Learned Building It
- Real systems are messy
- Deployment breaks things you didn't expect
- Debugging teaches more than courses
- AI is powerful when used in pipelines, not prompts
What's Next
This is just Phase 1.
I'm building:
- IOC extraction
- Personal threat tracking
- AI weekly briefings
- URL scanning tools
If You're Learning Cybersecurity
Don't just read.
Build something that:
- fetches
- processes
- alerts
- teaches
Because the fastest way to learn security…is to watch it happening live.
Links
Live: cybersec-news-agent.vercel.app Code: github.com/SoumyaKhaskel/cybersec-news-agent
Building > reading. Always.