Beyond the Agent · 4-weekend live bootcamp · December 2026 · 12 seats
AI agents do the volume now. Learn to find what they miss.
Four weekends, live with the founders. An AI agent hunts a full Application with planted vulnerabilities (a WarZone) first. You find what it missed.
One batch of 12. Seats close Monday, 30 November, or when they're gone.
- Early bird (till Friday, 20 Nov)
- ₹17,999
- Regular (21 to 30 Nov)
- ₹24,999
- GST
- + GST
- Seats
- 12. One batch.
- How you get in
- We review every seat. If we're unsure, we'll ask for a 15-minute call. If it's not the right level for you yet, a full refund within 48 hours, and we'll tell you what to practise first.
Who gets a seat
12 seats. Here's who they're for.
What you need
- You know the OWASP Top 10 beyond the names.
- You're comfortable in Burp or Caido.
- You can read an HTTP request and say what it does.
- No machine-learning background needed.
We review every seat ourselves, and may ask for a 15-minute call. It keeps the room at one level, so nobody waits and nobody drowns.
Why now
Agents are fast. Someone still has to check them.
AI agents now find a lot of vulnerabilities, and they're useful. But someone has to know what they missed, prove what's real, and steer them. That's the job this trains.
Weekend by weekend
What you'll do, weekend by weekend
Each day: about 1 hour of concepts, 3 hours hunting, then 1 hour of debrief on what you and the agent missed.
Weekend 1 · 5 to 6 Dec
The agent goes first.
An AI pentest agent runs a WarZone before you touch it. We show what it found and missed. Then you hunt the gaps it left.
Weekend 2 · 12 to 13 Dec
What agents miss.
Business logic, where nothing is technically broken. Access control across users and roles. Vulnerabilities that only appear when you chain steps.
Weekend 3 · 19 to 20 Dec
Trust, but verify.
Sort an agent's findings: drop false positives and duplicates. Prove what's real with evidence. Write the report.
Weekend 4 · 26 to 27 Dec, or 2 to 3 Jan
When AI is the target, then the final.
An AI WarZone where the AI itself is what you attack. Then a final WarGame on a fresh WarZone: you plus an agent, scored against the full answer.
Outcomes
What you leave with
- A public write-up, reviewed by the founders before you publish it.
- A repeatable way to run, steer and check AI hacking agents.
- Certificates you earn can be checked by anyone at verify.barracks.army.
- Top performers become eligible for a Synack Red Team referral from Barracks, based on their performance. Synack's wait-list and their own checks still apply.
Who teaches it
Taught live by both founders
Satyam Gothi (RogueSMG)
Co-founder and CEO
Former full-time bug bounty hunter on HackerOne and Bugcrowd. Former null Ahmedabad Chapter Lead. His YouTube channel appeared in Intigriti's 2020 "Top 20 bug bounty YouTube channels" list.
Kuldeep Pandya
Co-founder and CTO
Synack Red Team Hero for five consecutive years.
Frontier AI models run our WarZones first, so you see what a person finds that AI misses.
How it's different
How this differs from a typical AI-hacking course
- Focus
- Typical courseUsing AI tools to go faster
- Beyond the AgentKnowing what the AI missed, and proving it
- Feedback
- Typical courseDid you get the flag?
- Beyond the AgentA founder-reviewed write-up of what you found and what the agent missed
- Target
- Typical coursePuzzles and practice boxes
- Beyond the AgentFull Applications with planted vulnerabilities
- Format
- Typical courseRecorded
- Beyond the AgentLive, taught by both founders
FAQ
Questions
Which AI agent do we use?
We'll name the agent before weekend 1. It runs the same way for everyone.
Is it live or recorded?
Live.
Why only 12 seats?
Two founders, live, reading every finding and every report. Past 12, we'd be skimming your work instead of reading it.
What happens after I reserve?
We look at your profile ourselves. If your experience is clear, your seat is confirmed. If we're not sure, we'll set up a 15-minute call. If it's not the right level for you yet, you get a full refund within 48 hours and a short note on what to practise first.
Do I need machine-learning knowledge?
No. If you can test an Application with Burp or Caido, you can do this.
Will this get me a job?
Nobody can guarantee you a job. Not us, not a certification that costs lakhs. What we can make sure of: you leave able to do the part of the job agents can't, with proof you can show. A founder-reviewed write-up of what you found and what the agent missed, and a certificate.
How you get a seat
12 seats. Three steps.
Step 1
Apply
Tell us what you test. We reply with a payment link.
Step 2
We review
Maybe a 15-minute call to check your level. No trick questions.
Step 3
You're in
Or a full refund within 48 hours. Early bird till Friday, 20 Nov.
Seats close Monday, 30 November, or when all 12 are taken.