MindFort Announces $3M+ Seed to Secure the World's Software in the AI Era

MindFort, an applied AI security lab building autonomous agents that find, validate, and patch vulnerabilities in production software, today announced its $3M+ seed round, led by Soma Capital, with participation from Y Combinator, 468 Capital, CRV, Sandwith Ventures, and Blast.

The raise comes at an inflection point for software security. Anthropic's Project Glasswing and Claude Mythos Preview recently showed that frontier AI can autonomously uncover thousands of previously unknown vulnerabilities, including flaws that had survived decades of review. Attackers are already using the same AI to find and exploit bugs faster, and the past year's high-profile breaches have shown that yearly pentests and noisy scanners can't keep up with how quickly modern software ships. MindFort is building the autonomous security engineer for this new era: a production-ready system any company can deploy today.

"Software is being written faster than it can be secured, and the tools most teams rely on were not designed for that," said Brandon Veiseh, co-founder and CEO of MindFort.

MindFort's agents run autonomous pentests on a schedule or on every CI/CD push, chain vulnerabilities into validated findings with working proofs of exploit, and ship fixes as pull requests via native GitHub, Linear, and Jira integrations. Since launching, MindFort has signed enterprise customers and fast-growing startups, onboarding in under 15 minutes and delivering first results in hours instead of weeks. On internal and public benchmarks, its agents outperform legacy DAST, SAST, and automated scanners and have beaten human pentest teams head-to-head, with an extremely low false-positive rate and mean time to remediation measured in minutes.

Finding bugs is only half the problem. Most vulnerabilities don't live in application code alone; they live in cloud configuration, infrastructure, identity and access policies, network rules, and third-party integrations. Remediating across all of those surfaces, not just opening a code PR, is a core focus for MindFort, and the team is extending its agents to propose and apply fixes wherever the underlying issue actually lives.

Powering the platform is HillClimb, MindFort's recursive learning infrastructure that lets agents build a knowledge graph of each target and compound experience across every engagement. In parallel, the team is developing MF-1, a model purpose-built to bring frontier AI to offensive security and make security work both more performant and dramatically more affordable.

MindFort was founded by Brandon Veiseh and Akul Gupta. Veiseh, CEO, previously led product at ProjectDiscovery and most recently led AI tooling for offensive security at NetSPI. Gupta, CTO, leads MindFort's model and agent engineering; he has done frontier red-teaming for OpenAI and Anthropic, published at top AI and security conferences, and holds an M.S. in Computer Science from UIUC, where his research focused on applying LLMs to offensive security.

The seed funding will accelerate the growth of the company by funding an expansion of additional capacity to serve unprecedented demand, expand the applied AI research team to train MF-1, and broaden platform coverage across web, API, network, and cloud surfaces.

To learn more, visit mindfort.ai and request a demo today.

About MindFort:

MindFort's mission is to build the best AI security agents on earth to help automate cybersecurity and enable self-securing, self-healing organizations.