Mindfort deploys autonomous AI agents that continuously find, validate, and patch security vulnerabilities in web applications, operating as a 24/7 AI red team for engineering and security teams at startups and public companies. The launch marks MindFort's $3M+ seed announcement , which founder Brandon Veiseh says will fund infrastructure to meet growing demand and expand the team into new security verticals. It arrives as pressure mounts on companies to prepare defenses for AI-driven attacks, with quarterly manual pentests and noisy traditional scanners struggling to keep pace.
Teams point Mindfort at a live web app and thousands of agents run in parallel to surface exploitable vulnerabilities, then ship automated pull requests to fix them. The platform can run continuously, on a schedule, or inside CI, and its API exposes the same agents for tasks like triaging bug reports, validating findings, and running targeted security reviews. This launch also introduces Hill Climb, a recursive learning system that lets agents consolidate memories across runs so they become more efficient at finding real exploits over time, which addresses the false-positive problem where traditional scanners generate 30-50% false positives that typically drown engineering teams in noise.
Mindfort is a Y Combinator X25 company founded in 2025 by Akul Gupta and Brandon Veiseh , based in Los Angeles, with Sam Akhavan rounding out the founding team. Veiseh leads as CEO and Gupta as CTO, both with backgrounds spanning AI and cybersecurity. Soma Capital and Y Combinator are among the backers supporting the company's push to turn continuous, autonomous pentesting into a default part of the software delivery pipeline.
Comments (7)
Sign in to join the discussion.
P
Priya RavindranApr 23, 2026
Proving the exploit before shipping the fix is the correct order of operations. Half of security tooling skips step one and calls it 'detection'.
L
latentslothApr 23, 2026
Every pentest firm charging by the hour just felt a cold breeze.
K
Kosta ZervasApr 23, 2026
The landing page video pacing is weirdly good for a security launch. Most infra companies still think a terminal gif counts as a hero asset.
M
Mingfei WuApr 23, 2026
How autonomous are we talking when the agent hits a finding that needs codebase context it doesn't have? Curious where the human loop actually sits.
D
Dario OkaforApr 23, 2026
Autonomous agents plus compute-heavy exploit generation sounds like a gross margin conversation I'd want to have before scaling the team.
H
Hadil QureshiApr 23, 2026
Tweet copy is three bullets and a raise announcement with zero drama. Respect for not turning it into a 14-post thread.
V
Venkat RamanathanApr 23, 2026
The hard part isn't finding the vuln, it's the PR not breaking the build. Curious what the merge rate looks like on real repos.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Mindfort
Autonomous security agents that prove exploits — and ship fixes
Autonomous security agents that prove exploits — and ship fixes
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Mindfort deploys autonomous AI agents that continuously find, validate, and patch security vulnerabilities in web applications, operating as a 24/7 AI red team for engineering and security teams at startups and public companies. The launch marks MindFort's $3M+ seed announcement , which founder Brandon Veiseh says will fund infrastructure to meet growing demand and expand the team into new security verticals. It arrives as pressure mounts on companies to prepare defenses for AI-driven attacks, with quarterly manual pentests and noisy traditional scanners struggling to keep pace. Teams point Mindfort at a live web app and thousands of agents run in parallel to surface exploitable vulnerabilities, then ship automated pull requests to fix them. The platform can run continuously, on a schedule, or inside CI, and its API exposes the same agents for tasks like triaging bug reports, validating findings, and running targeted security reviews. This launch also introduces Hill Climb, a recursive learning system that lets agents consolidate memories across runs so they become more efficient at finding real exploits over time, which addresses the false-positive problem where traditional scanners generate 30-50% false positives that typically drown engineering teams in noise. Mindfort is a Y Combinator X25 company founded in 2025 by Akul Gupta and Brandon Veiseh , based in Los Angeles, with Sam Akhavan rounding out the founding team. Veiseh leads as CEO and Gupta as CTO, both with backgrounds spanning AI and cybersecurity. Soma Capital and Y Combinator are among the backers supporting the company's push to turn continuous, autonomous pentesting into a default part of the software delivery pipeline.
Comments (7)
Proving the exploit before shipping the fix is the correct order of operations. Half of security tooling skips step one and calls it 'detection'.
Every pentest firm charging by the hour just felt a cold breeze.
The landing page video pacing is weirdly good for a security launch. Most infra companies still think a terminal gif counts as a hero asset.
How autonomous are we talking when the agent hits a finding that needs codebase context it doesn't have? Curious where the human loop actually sits.
Autonomous agents plus compute-heavy exploit generation sounds like a gross margin conversation I'd want to have before scaling the team.
Tweet copy is three bullets and a raise announcement with zero drama. Respect for not turning it into a 14-post thread.
The hard part isn't finding the vuln, it's the PR not breaking the build. Curious what the merge rate looks like on real repos.