About Obsidian Security
Obsidian Security sells real-time visibility, continuous governance and runtime enforcement over agents, identities and integrations across third-party apps. Obsidian Security provides guardrails and real-time controls for AI agents that need access to third-party apps. The Obsidian platform turns billions of signals across business critical applications into security intelligence to protect users, agents, and data. Obsidian Security collects and connects third-party app and AI configurations, user activity, real-world threat signals, and browser telemetry.[3][4][13]
Obsidian Security says enterprise data lives in Salesforce, Workday, ServiceNow and 35K+ third-party enterprise apps that customers do not own. Obsidian Security says non-human identities now outnumber people inside the enterprise application estate. Obsidian Security says every AI application and agent connected to an enterprise inherits the permissions and trust misconfigurations that already existed in enterprise applications. Obsidian Security says vibe-coded workflows connect straight into Salesforce, Snowflake, Workday, and Slack, often before anyone has reviewed them. Obsidian Security spent years mapping every identity, permission, and data flow across third-party enterprise applications. Obsidian Security gives customers visibility into what is connected, controls to enforce what is allowed, and clarity on what needs to change.[3][5]
Obsidian Security has over 200 integrations into SaaS and AI applications that pull posture and activity data. The Obsidian browser extension captures real-time user and AI agent activity. Obsidian Security normalizes every human and non-human identity, entitlement, access, and action and connects them in a knowledge graph. The Obsidian Knowledge Graph maps the relationships between identities, permissions, and actions. Obsidian Security layers configuration, policy, and ownership data together to surface misconfigurations and privilege overlaps. Obsidian Security builds machine learning baselines from real enterprise behavior, refined by what it learns from actual incidents. Obsidian Security normalizes and correlates 29+ billion events from third-party applications every month and feeds them back into the platform. Obsidian Security builds detections from how attackers acted across 500+ real breach responses with DFIR firms. When one customer identifies and removes a risky integration or access pattern, Obsidian Security shares that signal automatically across every environment it protects. Obsidian Security provides out-of-the-box detection rules mapped to the MITRE ATT&CK framework and machine learning models based on real incident data. Obsidian Security integrates with SIEM and SOAR platforms for automated workflows. Obsidian Security provides an AI extension of the security team that explains rules, triages threats, and manages exceptions.[4][6][7][8][13]
Obsidian Security centralizes AI agent policy and enforces it at runtime across coding agents, agent platforms, MCP connections, and existing gateways. Obsidian Security agent policies use who the agent acts for, that person's employment status, and the team's data labels to tell a routine action from a risky one. Obsidian Security agent policies hold or block only the agent actions that match their conditions, and tool calls that match no policy proceed. Obsidian Security lets teams test each policy against real agent traffic before it stops anything. Obsidian Security says employees can connect AI agents to apps such as Salesforce, Snowflake, Slack, and Google Drive through MCP. The Obsidian proxy evaluates MCP traffic at each step of a session, including the app the agent can reach, the records it can list, and each tool call. The Obsidian proxy maps each MCP session to the person who started it, their account in the app, and the workspace, org, or instance the session reaches. With the Obsidian proxy, agents receive a proxy-issued token, and each user authorizes the app through the vendor's own consent flow.[5][9][10]
Obsidian Security monitors every new and existing MCP server in an environment with one dashboard. Obsidian Security shows MCP servers across AI agent platforms such as n8n and Microsoft Copilot. Obsidian Security raises risk alerts when agents connect to MCP servers without authentication. Obsidian Security helps teams spot unvetted or suspicious MCP servers and remove them.[11]
Obsidian Security maps supported identity, permission, configuration, and activity data to one control model. Obsidian Security provides mappings to NIST, CIS, ISO 27001, and GDPR. Obsidian Security continuously checks supported controls and shows what changed. Obsidian Security attaches the available identity, permission, and ownership context to each finding so teams can find the cause and route the fix. Obsidian Security routes fixes through Jira, ServiceNow, and other supported tools. Obsidian Security fixes misconfigurations, removes high-risk integrations, and stops malicious activity in third-party apps. Obsidian Security detects and removes risky or unauthorized AI agent actions in third-party apps. Obsidian Security pairs inline defenses against credential and session theft with alerts for investigation and response. Obsidian Security monitors configuration drift and spots violations to automate compliance reporting. Obsidian Security discovers app sprawl, including unsanctioned AI and shadow SaaS. Obsidian Security offers coverage for organisations of up to 1K users.[12][13][14]
Obsidian Security committed to securing identity and SaaS apps, which it describes as a new attack surface that nobody was watching. Security teams at large financial services and technology companies adopted Obsidian Security to detect account takeovers, SaaS threats and privilege misuse. Obsidian Security is backed by IVP, Norwest, and GV. Obsidian Security announced an $85M Series D led by Crescent Cove Advisors.[2][5][7][11][12][13]
Obsidian Security on the record