Comp AI has raised a $34 million Series A led by Roo Capital and Grand Ventures to scale its agentic platform for cybersecurity compliance and audit automation.
The startup addresses a specific revenue blocker: enterprise software companies often cannot close deals without SOC 2 certification, a labor-intensive compliance process. Comp AI's agents draft security policies, collect audit evidence, and continuously monitor control compliance, freeing engineering teams to focus on product work.
CEO Lewis Carhart said the business model is straightforward. "For a lot of software companies, security and compliance are directly tied to revenue," he said, citing customers who withhold contract signature until they receive a SOC 2 report. The platform also offers AI-powered penetration testing to identify vulnerabilities in code and infrastructure.
The founding team—Carhart, COO Claudio Fuentes, and CTO Mariano Fuentes—launched Comp AI in January 2024. Fuentes and his brother spent nearly a decade building startups together, most recently LeapAI, a workflow platform that reached over one million users before the founders shut it down. They concluded the product lacked a defensible use case.
That failure taught them two lessons: how to build with large language models and the importance of solving a specific, acute problem. The SOC 2 pain point became evident during LeapAI's attempts to serve larger enterprise customers. Fuentes described the certification process as "very obscure." He said, "It took us a couple of months of doing things by hand, and the whole time it meant taking our eyes off building the product."
Comp AI's agents automate the mechanical work while preserving human control. Carhart emphasized the guardrails: "An agent might draft a policy, for example, but a person still reviews and approves it." The company noted that as agents take on more consequential actions, the safeguards and human approval requirements should increase accordingly. The platform does not replace independent audits or human personnel.
