Exabeam Agent Sensor
The Exabeam Agent Sensor provides endpoint visibility for Agent Behavior Analytics (ABA). The sensor captures custom AI agent activities to fill visibility gaps left by standard platforms like Microsoft Copilot. It converts this data into standardized CIM events for downstream threat detection. The ABA platform includes two core components: An open-source library to track custom cloud agent actions from the inside and the Exabeam Agent Sensor deployed on endpoints to collect, normalize, and forward telemetry to Exabeam. You can export this data via webhooks or OpenTelemetry directly to downstream platforms like Dynatrace or Google Vertex AI.
The sensor provides visibility into AI agent usage across endpoints by capturing complete interaction flows with prompts and responses. The system records user, session, host, and model details and tracks tool usage, execution steps, latency, token usage, and cost. The architecture supports custom AI integrations through event hooks, while the Exabeam Agent Sensor collects, normalizes, and sends data to Exabeam for centralized analysis. Exabeam allows users to review interactions, monitor performance, analyze usage patterns, detect anomalies, and support compliance. You can search logs, investigate issues, and optimize AI workflows based on actionable insights. ABA enables you to quickly identify lateral movement, compromised accounts, and insider threats.
For more information and to get started with the sensor, see Exabeam Agent Sensor.
Key capabilities of the Exabeam Agent Sensor
End-to-End Interaction Visibility – Captures complete interaction flows between users and AI agents, including User prompts (inputs), and AI-generated responses (outputs).
Context-Rich Data Capture – Each interaction is enriched with critical context, such as user details, session interaction, host/device, model details. This contextual information helps for filtering, auditing, and troubleshooting.
Tool Usage and Execution Tracking – Tracks tools or integrations used during an interaction, and execution steps and outcomes. This helps understand how AI agents interact with other systems and perform tasks.
Performance and Cost Metrics – Captures important operational metrics, including latency or response time for each interaction, volume of input/output processed, and estimated or actual usage cost. These insights help monitor performance and control resource usage.
Compliance and governance – Ensure responsible and secure AI usage.
Troubleshooting support – Quickly debug issues with full interaction context.
Search Events
Exabeam Agent Sensor ingests and parses high-volume cloud logs to generate structured behavioral insights for users and systems.
Analyze Prompt and Response Flows
Exabeam Agent Sensor ingests and parses high-volume cloud logs to generate structured behavioral insights for users and systems. Each interaction is captured with context such as session, user, host, and model details.
Monitor Usage Patterns and Activity Across Users and Endpoints
This AI-driven detection engine enables you to identify lateral movement, account compromise, and insider threats.