Use Cases
Overview
Cerbera AI addresses a set of risks that emerged with the rapid adoption of AI tools and agents, and that existing tooling (EDR, firewalls, VPNs) was not designed to catch. Each use case below pairs the problem with how Cerbera AI handles it.
Discover Shadow AI
Problem. You do not know which AI tools, agents, and MCP servers your employees use, or in what proportion.
How Cerbera AI helps. Deployed in monitor-only mode, the proxy inventories every AI tool, model, agent, and MCP server in use across the fleet, with no user impact. You get a census before you set any policy.
See AI Discovery.
Block Unsanctioned Tools
Problem. Employees install tools you have not vetted, sometimes intrusive clients like OpenClaw or models like Deepseek, on professional machines.
How Cerbera AI helps. Approve sanctioned tools and block the rest with firewall-style rules. A blocked user sees a pop-up explaining why and can request an exception.
See Rules and Exceptions & Remediation.
Prevent Secret Leakage
Problem. Prompts and agent requests routinely contain API keys, AWS keys, tokens, or PII that should never reach an AI provider.
How Cerbera AI helps. Redact rules strip secrets from a request before it leaves the device. The workflow continues uninterrupted, and you avoid having to rotate exposed credentials.
See Rules.
Stop Personal Account Usage
Problem. Employees use personal accounts on Claude Code, ChatGPT, and similar tools. When they leave, there is no way to delete the company data in those sessions or revoke access, and your org-wide security settings do not apply.
How Cerbera AI helps. Detect non-enterprise account usage and require enterprise accounts through alerts and rules.
See Agent Controls.
Constrain Risky Agent Behavior
Problem. AI agents do things a human user would not: opening SSH sessions, reading .env files, or connecting to a production database.
How Cerbera AI helps. Behavior-based rules flag or block these actions, which an EDR (designed around human usage) may let through.
See Agent Controls.
Govern MCP Servers
Problem. Employees connect to third-party MCP servers maintained by unknown parties, often handing over API keys and tokens.
How Cerbera AI helps. Inventory every MCP server, flag token exposure and misconfiguration, and enforce allow or block policies, including a deny-by-default posture once the legitimate servers are known.
See MCP Governance.
Produce AI Audit Evidence
Problem. Auditors increasingly expect evidence of how AI tools are used and governed inside the organization.
How Cerbera AI helps. Alerts and activity export in OpenTelemetry format and provide an audit trail of AI usage and policy enforcement for frameworks like SOC 2 and ISO 27001.
See Dashboards & Alerts and Openness & Interoperability.
Catch Supply-Chain Risk (Roadmap)
Problem. Coding agents pull open-source packages for just-in-time scripts that never pass through a CI/CD pipeline, so nothing filters them.
How Cerbera AI helps. Cerbera is exploring analysis of packages an agent downloads for one-off scripts, checking them for known vulnerabilities. This is on the roadmap.
See Agent Controls.