iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Opsin announced on October 15, 2026, that it raised $21 million in a round led by SYN Ventures and Race Capital. The enterprise AI security company says it will use the funding to accelerate product development, expand in regulated industries, strengthen its risk-protection layer and hire staff.
Opsin’s platform is designed to answer three operational questions security teams increasingly face: which AI agents are active, what systems and data they can reach, and whether their runtime behavior matches their intended purpose.
What the funding means
Opsin was founded by James Pham, co-founder and CEO; Oz Wasserman, co-founder and chief product officer; and Jeremy Mailen, co-founder and chief technology officer. The company positions itself as an enterprise AI security and governance provider for organizations ranging from mid-sized businesses to Fortune 500 companies, particularly in healthcare, financial services, manufacturing and other regulated sectors.
Free tools Windows power users keep installed
One-click scans. No signup required.
The new capital is earmarked for faster product development, broader coverage of regulated-industry use cases, continued investment in Opsin’s risk-protection layer and hiring. The announcement names Barry-Wehmiller, Culligan, UiPath and Wellstar Health System in its customer or reference material. Those customer references are company-provided and do not constitute an independent assessment of the product.
What Opsin’s platform monitors
Opsin says it builds a continuously updated inventory of enterprise AI agents. For each discovered agent, the inventory can show:
- the person or team responsible for the agent;
- when the agent was last active;
- the permissions and delegated credentials it has;
- the tools and services it can invoke; and
- the data and systems within its potential reach.
Discovery uses delegated API access across supported enterprise AI platforms. Opsin says this approach does not require endpoint software, browser extensions or employee surveys. Its materials describe normalization across Microsoft 365 Copilot, Copilot Studio, Claude Enterprise, ChatGPT Enterprise and Gemini, allowing security teams to examine agents across multiple platforms in a common view.
How the Risk Graph identifies dangerous agents
The central analysis layer is Opsin’s Risk Graph. It links an agent’s stated intent, identity, permissions, data access, tools and observed behavior. The aim is to distinguish an agent that merely has broad theoretical access from one whose access and actions create a meaningful risk.
Opsin compares what an agent was intended to do with what it can reach and how it behaves during operation. That supports several investigations:
| Security question | Evidence Opsin says it provides |
|---|---|
| Which AI agents are active? | A current inventory with activity information, ownership and platform context. |
| What data can an agent access? | Mapped permissions, connected tools and data reach. |
| How do I find over-permissioned agents? | A comparison between intended duties and the systems, data and actions available to the agent. |
| Is an agent acting outside its purpose? | Observed behavior evaluated against stated intent and policy boundaries. |
When a boundary is crossed, Opsin says a team can notify the owner, block the action or quarantine the agent. The model is evidence-backed prioritization and remediation rather than automatically banning every AI deployment.
Why agent visibility has become a governance problem
Opsin’s homepage cites the following market figures, with the original attribution shown by the company:
| Figure | Attribution | What it indicates |
|---|---|---|
| 60% of the workforce has access to sanctioned AI tools | Deloitte, 2026 | Approved tools are already broadly available to employees. |
| 82% of enterprises have unknown AI agents across their environments | Cloud Security Alliance, 2026 | Organizations may not know which agents exist or who owns them. |
| 74% report over-permissioned agents | Cloud Security Alliance, 2026 | Many organizations see agents with more access than their stated duties require. |
| 79% report hard-to-monitor access pathways created by agents | Cloud Security Alliance, 2026 | Agent-mediated connections can be difficult to observe with conventional controls. |
These figures describe the market context, not an independent audit of Opsin. Opsin separately reports 67% faster investigation and 60% less exposure as company-stated impact outcomes; those percentages have not been presented as independently verified study results.
What an investigation could look like
- Establish the inventory. Security staff review the agents discovered through delegated connections and confirm owners, activity and platform.
- Trace reach. They inspect the agent’s permissions, tools, connected services and accessible data.
- Compare intent with capability. The team checks whether the access granted is necessary for the agent’s documented job.
- Review behavior. Runtime actions are evaluated for activity that departs from the intended purpose or violates a policy boundary.
- Choose a response. Depending on the evidence, the team can contact the owner, block a specific action, quarantine the agent or remediate its permissions.
This workflow is intended to let organizations keep useful automation running while concentrating controls on agents whose ownership, access or behavior creates a demonstrable problem.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Customer perspective and market positioning
Opsin’s published customer statements emphasize context rather than a simple allow-or-deny decision. Lior Eldad, vice president of IT at UiPath, said: “I need to know whether an agent’s connection to sensitive systems is risky or legitimate. Opsin gives me the deep context to tell the difference.” Jim Shelby, global IT security engineer at Culligan, said: “Finally, I have a way to assess agent risks based on its capabilities and whether those capabilities align with the intended purposes.”
Those statements illustrate the platform’s intended role: connecting identity, access and behavior so a security team can explain why an agent is risky and select a proportionate response. They are testimonials supplied by Opsin, not controlled comparative tests.
What the announcement does not establish
- No independent study in the announcement verifies Opsin’s 67% investigation-speed or 60% exposure-reduction figures.
- The materials describe an API-connected enterprise software platform, not a physical appliance or endpoint product.
- Coverage depends on the enterprise AI platforms and delegated connections an organization can authorize; the named integrations are Microsoft 365 Copilot, Copilot Studio, Claude Enterprise, ChatGPT Enterprise and Gemini.
- Customer names and industry coverage demonstrate the company’s stated market focus, but they do not by themselves prove performance for every deployment.
Why the round matters for enterprise buyers
AI agents can combine a user’s identity, long-lived permissions and automated tool calls. Traditional inventories may show applications or accounts without revealing the agent’s purpose, current activity or path to sensitive data. Opsin’s approach treats the agent as the unit of governance and joins those signals in one risk model.
For security leaders, the practical value is a path from discovery to action: identify previously unknown agents, assign accountability, narrow excessive access, investigate behavior that departs from intent and document why an action was allowed, blocked or quarantined. The funding gives Opsin additional resources to develop that model and pursue more regulated-industry deployments.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
