When a security team at a hospital or a manufacturer signs an enterprise-wide contract for Microsoft Copilot or ChatGPT Enterprise, they are not just buying a chatbot. They are provisioning an unknown number of semi-autonomous agents, each with its own potential access to patient records, financial data, and proprietary designs. The procurement cycle ends, but the governance problem has just begun. Opsin, a San Jose-based startup, is betting that the budget for solving that problem sits with the CISO, not the head of AI.
Founded in 2024, Opsin has raised $28 million across two rounds, including a $21 million seed in October 2026 led by SYN Ventures and Race Capital [Business Insider, October 2026]. The company reports serving nearly 20 enterprise customers in regulated sectors like healthcare and manufacturing, including Wellstar Health System, Barry-Wehmiller, and UiPath [Morningstar/Accesswire, October 2026]. Its pitch is a platform that discovers, monitors, and governs AI agents across approved tools, providing a unified view of what these agents can access and what they are actually doing [PERPLEXITY SONAR PRO BRIEF]. For a team of roughly 20 people, that is a focused, high-stakes wedge into a market that is still defining its own perimeter.
The Wedge: Runtime Security for Autonomous Agents
Opsin's initial product focus was on preventing sensitive-data oversharing by generative AI tools, a clear and immediate pain point for any company that has rolled out Copilot [PERPLEXITY SONAR PRO BRIEF]. The company has since expanded its positioning to cover the full lifecycle of an AI agent, from discovery and identity to continuous behavioral monitoring and risk assessment. The core of the offering is context: Opsin claims to combine identity, agent, data, tool, behavioral, and intent context to show security teams what an agent is supposed to do, and flag when its behavior changes [PERPLEXITY SONAR PRO BRIEF].
This moves the conversation from static policy setting to runtime security. Instead of just blocking or allowing access, the platform is designed to evaluate agents in context, comparing their stated intent, permissions, and actual activity to surface risk signals [Opsin | AI Agent Security Platform]. For a customer like Wellstar Health System, Opsin delivered a risk assessment within three days for petabytes of data across hundreds of Microsoft Teams and SharePoint sites [Opsin Customer Stories | Wellstar Health System, 2026]. That kind of speed is a tangible proof point for an enterprise buyer under pressure.
A Team Built for the Enterprise Sale
The founding team carries credentials that resonate in a security procurement meeting. CEO James Pham and CPO Oz Wasserman both worked at email security company Abnormal Security, leaving in 2022 [PERPLEXITY SONAR PRO BRIEF]. Pham holds an MBA from MIT and focused on data analytics and AI [Opsin Security Blog]. Wasserman is a cybersecurity veteran with prior roles at FireEye and Reco AI [ZoomInfo]. CTO Jeremy Mailen was previously VP of Engineering at Exterro and co-founded Divebell, a data security platform that was acquired by Exterro [Highperformr, 2026]. This blend of AI, security, and prior startup experience is precisely the background an investor would want to see before writing a check for a governance platform.
The company's early hiring reinforces its go-to-market focus. Open roles have included a Cybersecurity Senior Account Executive for the East and Central US and a Senior AI Engineer in San Diego [PERPLEXITY SONAR PRO BRIEF]. This suggests a disciplined build-out, prioritizing sales to land more enterprise logos and engineering to deepen the product's technical moat.
| Role | Name | Prior Experience |
|---|---|---|
| Co-Founder & CEO | James Pham | Abnormal Security, MIT MBA |
| Co-Founder & CPO | Oz Wasserman | Abnormal Security, FireEye, Reco AI |
| Co-Founder & CTO | Jeremy Mailen | Exterro, Divebell (acquired by Exterro) |
Traction in the Regulated Core
Opsin's reported customer list is a strategic asset. Landing names like Wellstar Health System (healthcare), Barry-Wehmiller (manufacturing), and Culligan (industrial) does more than just bring in revenue. It provides case studies in highly regulated environments where data governance is non-negotiable and compliance frameworks like HIPAA, GDPR, and the EU AI Act are daily concerns [AI Governance Explained: Framework, Risks & Enterprise Controls]. These early customers become referenceable proof that Opsin's platform can handle complex, sensitive environments.
The company's reported $21 million seed round, led by specialized cybersecurity investor SYN Ventures and early-stage firm Race Capital, is a significant vote of confidence for a company at this stage [Business Insider, October 2026]. The capital is earmarked for expanding sales and engineering teams, a clear signal that the immediate plan is to scale go-to-market efforts and product development in parallel.
The Realistic Competitive Set
No direct competitors are named in public sources, but the realistic competitive set for Opsin is multi-layered. It is not competing with the AI platforms themselves (Microsoft, Google, Anthropic). Instead, it is vying for budget and attention within the security stack.
- Legacy CSPM and DSPM vendors. Cloud security posture management and data security posture management tools have broad visibility but are not architected for the dynamic, intent-based behavior of AI agents. Their wedge is infrastructure, not agentic workflows.
- AI safety and evaluation startups. A growing cohort of companies focus on pre-deployment red-teaming and output evaluation. Opsin's differentiation is continuous, runtime monitoring integrated into the agent's operational environment.
- Internal builds. Large enterprises with mature security teams may initially try to build monitoring in-house. Opsin's bet is that the complexity and speed of agent deployment will make a dedicated, integrated platform the more efficient choice.
The ideal customer profile is clear: a CISO or head of cloud security at a company with 1,000+ employees that has already standardized on one or more enterprise AI platforms (Copilot, ChatGPT Enterprise, Claude) and is now grappling with the operational and compliance risks of agent proliferation. This buyer has a budget for tools that reduce mean time to detection and response, and they need to report to auditors. Opsin is selling into that procurement cycle with a product built for runtime, not just setup.
What to Watch in the Next Twelve Months
The next year will be about proving that initial wedge can drive expansion. Key milestones to watch will be whether Opsin can:
- Convert its nearly 20 initial customers into multi-year commitments with growing seat counts or data volume.
- Land a marquee financial services or pharmaceutical logo, sectors with equally stringent data governance needs.
- Demonstrate clear product-led growth signals, such as usage-based expansion within existing accounts.
The $21 million war chest provides runway, but the market is moving quickly. The company's ability to stay ahead of both the AI platforms' native features and the inevitable competitive entrants will depend on execution velocity. For now, Opsin has secured funding, a credible team, and a foothold in the exact customer segment that needs its product most.
Sources
- [Business Insider, October 2026] This startup raised $21 million to clean up the AI agent mess hitting companies | https://africa.businessinsider.com/news/this-startup-raised-dollar21-million-to-clean-up-the-ai-agent-mess-hitting-companies/37e80nm
- [Morningstar/Accesswire, October 2026] Opsin Raises $21M to Keep AI Agents on Task and Out of Trouble | https://www.morningstar.com/news/accesswire/1236364msn/opsin-raises-21m-to-keep-ai-agents-on-task-and-out-of-trouble
- [PERPLEXITY SONAR PRO BRIEF] Product, team, and funding overview
- [Opsin | AI Agent Security Platform] Product description and capabilities
- [Opsin Customer Stories | Wellstar Health System, 2026] Wellstar Health System case study
- [AI Governance Explained: Framework, Risks & Enterprise Controls] Compliance frameworks mention
- [Highperformr, 2026] Jeremy Mailen background
- [ZoomInfo] Oz Wasserman background
- [Opsin Security Blog] James Pham educational background