The cybersecurity company is now valued at $1.1 billion following the funding round, which was led by Crescent Cove Advisors.
Obsidian positions its platform as an AI-first security solution designed to help businesses govern data, artificial intelligence systems and autonomous agents operating within third-party applications.
According to CEO Hasan Imam, the company wants to become a key security layer for organisations as AI agents gain greater access to business systems and sensitive information.
“We intend to be the platform that protects enterprises from agents going rogue in third-party applications,” Imam said.
The rapid adoption of AI agents is creating a growing challenge for security teams. Unlike traditional software, AI agents can independently perform tasks, access information and interact with multiple applications.
This can create significant risks when agents are given permissions that are broader than necessary.
Obsidian says more than 70% of its customers already allow AI agents to operate inside third-party applications, highlighting how quickly the technology is becoming embedded in business operations.
The company says its platform provides organisations with greater visibility into what AI agents can access and allows security teams to restrict potentially dangerous activity.
Its technology is designed to detect issues such as excessive data access, privilege escalation and policy violations while agents are operating.
Alongside the funding announcement, Obsidian unveiled several new AI security capabilities.
These include Agent Access Governance for Anthropic’s Claude Code and Cowork, runtime protection for AI agents, an inventory of Model Context Protocol (MCP) servers and an LLM inventory designed to help organisations track the models powering their AI agents.
The company says its runtime protection can work with agents built on platforms including Microsoft Copilot, Anthropic Claude, Google Vertex, Amazon Bedrock and OpenAI.
The technology is intended to stop potentially destructive actions before they can affect business systems.
The expansion of agentic AI has raised concerns about what could happen when autonomous systems are given access to sensitive business environments.
Obsidian points to examples involving AI agents triggering prolonged cloud outages, deleting important files and bypassing application safeguards to access production databases and backups.
For enterprises, the challenge is therefore no longer simply deciding whether to adopt AI. It is increasingly about determining what AI agents are allowed to do once they are inside corporate systems.
Jun Hong Heng, chief investment officer at Crescent Cove Advisors, said the adoption of AI will increasingly depend on trust, visibility and control.
“AI is fundamentally reshaping how businesses operate but their adoption will depend on trust, visibility and control,” he said.
Obsidian claims its technology is already used by 60 Fortune 500 companies, including organisations in financial services, telecommunications and social media.
As businesses continue deploying increasingly autonomous AI systems, the company is betting that controlling these digital workers will become just as important as deploying them.
Main Image: Finimize
