AI Agents That Find, Prove and Fix Security Bugs in Code
Security researchers turned founders: AI agents check every code change for exploitable bugs, show proof and suggest a fix.
In plain English
Hacktron uses AI to look for security holes in software before attackers do. Every time a team changes its code, the AI checks it, tries to prove a real bug exists, and suggests a fix. This catches problems a busy team might miss. Companies pay for ongoing testing.
~$240K revenue in first nine months (Cambridge Enterprise release, May 2026)
Operator: Zayne Zhang (CEO), Mohan Pedhapati (CTO), Harsh Jaiswal (research)
A named, real operator ran this and shared results publicly. The figures are their own.
Not disclosed.
The tool stack
Payments
Anthropic API ↗LLM / agents
Next.js ↗Web app framework
Prices are the vendors' published list prices as researched in Oct 2026 and change often — always check the vendor's pricing page.
Step-by-step setup
- Start from deep security know-how; the founders were known bug hunters.
- Run AI agents on each code change to look for exploitable bugs.
- Only report bugs the agent can prove, with a suggested fix.
- Sell yearly plans to software teams; grow via research write-ups.
What's automated vs. the human 20%
The machine does
- Core AI generation/analysis
- Billing and onboarding
You still do
- Product quality and support
- Distribution and marketing
- Compliance and edge cases
Overview
Security researchers turned founders: AI agents check every code change for exploitable bugs, show proof and suggest a fix.
Operator: Zayne Zhang (CEO), Mohan Pedhapati (CTO), Harsh Jaiswal (research). Website: https://www.hacktron.ai/
Estimated split: about 80% of the work automated, 20% human.
Realistic income
Cambridge Enterprise, which invested in Hacktron’s $2.9M pre-seed round, wrote in May 2026 that the company made about $240,000 in revenue in its first nine months. This figure is operator/investor-reported, not independently verified. This is one operator’s result, often after months or years of work, and doesn’t predict yours. Check the live source — numbers change.
Risks
- The ~$240K figure is from an investor’s news release.
- False alarms waste engineers’ time and erode trust.
- Security tools face high expectations and liability.
- Startup cost and time to first dollar are not disclosed.
*Income figures are illustrative ranges based on public pricing and typical outcomes reported by practitioners, not guarantees. Most new online businesses earn little or nothing at first. Not financial, legal or tax advice.
THE DAILY 6
The 6 best AI money stacks, in your inbox every morning.
Each pick comes with its tools and first steps, so you can start one in minutes.
Unsubscribe anytime.
Related stacks
SaaS
Adversarial AI Red-Teaming Infrastructure
LLM Infrastructure for adversarial, high-risk scenarios. Our adversarial AI helps authorised testers to simulate red teaming drills.
- Automation
- 85%
- Difficulty
- Hard
- Startup
- Not disclosed
$40,278 Last 30 days
$40,278 verified revenue last 30 days, mostly non-recurring (TrustMRR/Stripe, Oct 28, 2026)
VerifiedSaaS
Paid Library of Agent Skills for Claude Code & Cursor
Skillry is a growing library of Agent Skills for Claude Code, Codex, Cursor and other AI agents. Preview real outputs for Skills that create websites,…
- Automation
- 85%
- Difficulty
- Hard
- Startup
- Not disclosed
$3,918 MRR
~$3,918 MRR; $12,037 verified revenue last 30 days (TrustMRR/Stripe, Oct 27, 2026)
VerifiedSaaS
Agentic Back Office for Construction & Installers
Agentic AI SaaS automating back-office work for construction and installation companies, covering quoting, invoice reconciliation, budget control and…
- Automation
- 75%
- Difficulty
- Hard
- Startup
- Not disclosed
$400 MRR
~$400 MRR; $485 verified revenue last 30 days (TrustMRR/Stripe, Oct 25, 2026)
VerifiedBrowse more: AI developer tool business ideas