When an AI agent can write code, update software, or work across company systems, it may need access to the same sensitive tools employees use. That creates an important question: how do you let an agent get the job done without handing it the keys to the company?
Infisical is betting that answering that question is becoming a major business in its own right. On October 27, the security infrastructure company announced a $40 million Series B at a $600 million valuation. The round was led by Elad Gil, with participation from Gradient Ventures and Y Combinator, among others.
The funding is a notable milestone for the company, but the larger story is the problem it wants to solve. As AI agents move from experiments into everyday workflows, companies need ways to grant them access without exposing the passwords, API keys, and other credentials that protect critical systems.
The quiet risk behind an AI agent’s access
A company’s systems rely on secrets: credentials that allow software, services, and people to connect to one another. They can include an API key that lets an application use a service, a certificate that verifies a connection, or an encryption key that protects sensitive data.
Traditionally, teams have built processes around deciding who or what can use those secrets, and when. But AI agents add a new kind of user to the mix. They can take actions across tools and systems, sometimes at a speed and scale that make old, manual approaches harder to manage.
Giving an agent a credential directly can create risk. If it is exposed, stored in the wrong place, or used beyond its intended task, it may open a path to sensitive systems. Yet keeping agents locked out entirely can undermine the very work companies hope to automate.
That tension, enabling useful access while limiting exposure, is the business opportunity Infisical is pursuing.
What Infisical is building
Founded by Vlad Matsiiako, Maidul Islam, and Tony (Tuan) Dang, Infisical provides infrastructure for managing secrets, certificates, encryption keys, and privileged access. The company is open source and a Y Combinator alum.
Its next area of investment is Agent Vault, a feature designed to let AI agents interact with sensitive services through a proxy. Rather than giving an agent the actual credential, the system acts as an intermediary: the agent makes a request, and the secure infrastructure helps carry out the authorized interaction without revealing the underlying secret.
That distinction matters. An agent may need to perform a task, but it does not necessarily need to know the password or API key that makes the task possible. The goal is to separate permission to do something from possession of the credential that grants access.
Infisical says more than 1,300 organizations use its platform. The new funding will support further product development, including Agent Vault, as well as engineering and go-to-market growth. The company’s own announcement also says it has been cash-flow positive, giving it room to invest in the opportunity from a position of strength.
Why investors are paying attention now
AI agents are moving into a landscape that existing security systems were not always designed to handle. Companies are using more software, connecting more services, and experimenting with automated tools that can act across them. Each new connection can raise a practical question: what should be allowed, and how can that access be controlled?
That makes the “trust layer” around AI more than a technical detail. It is part of what could make the technology usable in organizations where a mistake can expose customer data, disrupt operations, or create costly security problems.
For entrepreneurs, there is a broader lesson here: a new technology wave can create markets not only for the headline product, but also for the infrastructure that makes the product safer and easier to adopt. Cloud computing created demand for new ways to secure and manage cloud systems. AI agents are creating fresh questions about identity, access, oversight, and accountability.
The companies that notice those questions early may have an opportunity to define how an emerging category works. That does not guarantee success: security products still need to earn trust, fit into customers’ workflows, and prove their value. But it does explain why a company focused on the less-visible foundation of AI can attract significant attention.
The business case for making security feel usable
Security can sometimes be treated as a brake on progress: another approval, another policy, another thing teams have to work around. But when protection is difficult to use, people may look for shortcuts. That can leave sensitive systems less secure, not more.
Infisical’s bet is that access controls should keep pace with the way developers and organizations actually work. For AI agents, that means designing permissions around specific tasks instead of relying on broad, long-lived credentials.
For a business considering AI agents, the practical questions are already useful to ask:
- What systems would an agent need to access to complete a real task?
- Can that access be limited to only what the task requires?
- Is the credential hidden from the agent, and can access be revoked?
- Is there a clear record of what the agent did?
Those questions can help organizations move beyond a simple yes-or-no decision about automation. They create a path toward trying AI in a controlled way, with attention to the boundaries that matter.
And for founders, the same questions can be a way to find a market. Look closely at the workflows customers are eager to automate. Then look at what makes them hesitate: unclear permissions, poor visibility, fear of mistakes, or an inability to undo an action. Those points of friction may reveal a product opportunity.
Watch what breaks as AI spreads
Infisical’s $40 million round is a reminder that a technology’s supporting infrastructure can become a compelling category of its own. The agent may be the visible innovation, but the systems that let it act responsibly may determine whether companies are willing to use it.
That creates a timely challenge for builders: pay attention to the places where adoption meets risk. When a new tool starts handling work once reserved for people, ask what needs to change around it, not just to make it faster, but to make it trustworthy.
The next generation of successful businesses may come from solving precisely those problems. When a new wave of technology creates a new class of risk, the safest business can be the one that makes the wave usable.
So, whether you are building a company, creating a product, or choosing where to focus your skills, watch what breaks as AI agents spread through organizations. The breaking point may be more than a warning sign. It could be the start of a market.