Infisical raises $40M at $600M valuation to secure AI agent access

Infisical has raised $40 million in a Series B round at a $600 million valuation, led by Elad Gil, to build security infrastructure that lets AI agents work inside enterprise systems without exposing credentials. The round includes existing investors Gradient and Y Combinator, and brings the company’s total funding to an undisclosed amount.

The funding arrives as enterprises adopt AI agents for tasks like incident investigation, code review, and operational automation. These agents often need access to repositories, cloud services, and business applications, but traditional methods of embedding credentials create vulnerabilities. Infisical’s Agent Vault addresses this by acting as a forward proxy, ensuring sensitive credentials never enter the agent’s environment. Instead, the proxy handles authentication transparently, keeping secrets isolated from the agent’s runtime and logs.

This design reduces exposure, but it doesn’t eliminate all risks. Even with credential controls in place, granting an agent permissions to interact with systems requires careful scoping of what actions it can perform. Infisical’s approach includes access bundles and time-bound sessions to help enterprises define these boundaries, along with encrypted logs for auditing.

Infisical already serves over 1,300 customers and 300,000 developers, securing 10 billion secrets daily, and its open-source core has earned more than 30,000 GitHub stars. Named customers include Airbus, Volkswagen, Hugging Face, and Wise, though the company hasn’t specified how many use Agent Vault. The new funding will support further development of its agent security offerings while maintaining its broader platform for secrets management, certificates, and privileged access.

The round underscores a growing focus on AI agent security in enterprises. Earlier coverage highlighted startups like Island and Armadin tackling different aspects of this challenge—browser isolation and broader AI security tools, respectively. Infisical’s focus on credential isolation suggests a belief that this layer will become critical as agents take on more operational roles. The company’s hiring push—over 100 roles across engineering, sales, and product—reflects confidence in both its existing business and its expansion into agent access.

Elad Gil, who also led Infisical’s Series A, positioned the company as foundational for the AI era. His comments framed AI agents as a new category of enterprise users requiring dedicated security measures, not just incremental improvements to existing systems.

The next question is whether Infisical can scale Agent Vault beyond its preview stage and establish it as a standard. Its existing traction and open-source adoption provide momentum, but the AI agent security market remains nascent. Competitors are exploring different angles, and Infisical’s bet is that credential isolation will prove essential. If successful, it could shape how enterprises balance delegation and control as AI agents become more integrated into workflows.

Sources: unite.ai