Liminal Secures $5.5M Funding Led by OM Venture Capital to Enhance GenAI Solutions with Data Security Focus

Liminal just pulled in $5.5M, and this one feels different. Not because capital is rare. Capital chases gravity. This is about where gravity is forming in AI security. Liminal, the generative AI security and governance platform built for regulated enterprises, secured the round led by OM Venture Capital, with Florida Funders, Draper Associates, Framework Venture Partners, and High Alpha stepping in again. That is not tourist money. That is conviction money.

Respect to Founder and CEO Steven Walchek and Co-Founder and CTO Aaron Bach. 2 operators who saw something most enterprises were whispering about but not solving. Generative AI was moving fast. Security teams were moving carefully. Productivity was stuck in the middle, waiting for permission. Liminal decided to live in that middle.

Unlimited multi-model access wrapped in enterprise-grade governance. OpenAI, Anthropic, Google, Perplexity. Bring your own models. Detect sensitive data before it leaves the building. Protect it with policy. Rehydrate it after inference so employees get value without exposing PHI, PII, PCI, or intellectual property. It is not hype. It is plumbing. The kind that keeps the house from flooding while everyone is busy admiring the smart fridge.

CBO Michelle Eatherton and CMO Marc Jacocks are helping translate that into market motion. Because selling security in the age of AI is not about fear. It is about enablement. It is about giving CISOs and CIOs a way to say yes without losing sleep.

And the traction is not theoretical. Delta Dental of New Jersey and Connecticut reports a 5x ROI and 5x time improvement in AI workflows. Northern Arizona Healthcare saw 86,000 sensitive data elements detected and protected in just months. That is what happens when governance is not a speed bump but a seatbelt.

The lesson for founders watching this round is simple. Do not build the shiny object. Build the control layer everyone realizes they needed 6 months too late. Liminal did not try to out-model the model companies. They built the layer that makes every model enterprise-ready.

In a world where thousands of generative AI models and over 12,000 AI-enabled applications have flooded the market in under 2 years, enterprises are not asking which model is smartest. They are asking which deployment is safest.