# MindFort

> MindFort is a San Francisco AI security lab building autonomous agents that continuously find, validate, and patch vulnerabilities in web applications and APIs. Founded out of Y Combinator's X25 batch by offensive-security veterans, it runs like a 24/7 AI red team - proving exploits in isolated runtime environments before it reports them, then opening pull requests with the fix. The company raised a $3M+ seed led by Soma Capital in April 2026.

- **Founded:** 2025
- **Headquarters:** San Francisco, California, United States
- **Founders:** Brandon Veiseh (Co-founder and CEO), Akul Gupta (Co-founder and CTO), Sam (Co-founder)
- **Team size:** ~7 employees
- **Products:** Continuous Pentesting, Exploit Validation, Automated Patching, PR Security Review, MF-1
- **Notable:** Raised $3M+ seed led by Soma Capital (April 2026)., Graduated from Y Combinator X25 batch., Built proprietary offensive-security model (MF-1) and recursive learning infrastructure (HillClimb).

## Products & services

- **Continuous Pentesting** — Autonomous agents run 24/7 or on CI/CD triggers, mapping the attack surface and delivering proof-of-concept exploits rather than raw alerts.
- **Exploit Validation** — Every potential vulnerability is confirmed by exploiting it in an isolated runtime environment before it is reported, keeping false positives under 1%.
- **Automated Patching** — Agents analyze the codebase and open pull requests with contextual fixes, then re-test to confirm the fix holds.
- **PR Security Review** — Security analysis integrated into the development workflow, reviewing pull requests as they are opened.
- **MF-1** — A model purpose-built for offensive-security applications, tuned to reason about attack surfaces the way a human red-teamer would.
- **HillClimb** — Recursive learning infrastructure that builds a knowledge graph and compounds experience across engagements so each test makes the next smarter.

## Achievements

- Raised $3M+ seed led by Soma Capital (April 2026).
- Graduated from Y Combinator X25 batch.
- Built proprietary offensive-security model (MF-1) and recursive learning infrastructure (HillClimb).
- Reports under 1% false-positive rate by validating exploits before reporting.
- Agents reported to find vulnerabilities missed by existing tools and to score well on penetration-testing benchmarks.

## Latest updates

- **2026-04** — Announced $3M+ seed round led by Soma Capital to secure the world's software in the AI era.
- **2025-05** — Launched on Hacker News as MindFort (YC X25) - AI agents for continuous pentesting.

## Links

- Website: https://mindfort.ai
- LinkedIn: https://www.linkedin.com/company/mindfort-ai

---

Profile page: https://yespress.io/mindfort-ai-yc-x25
Published by YesPress — https://yespress.io
Last updated: 2026-07-30