Why AI Agents Are Blowing Up Your Telemetry Bill with Nikhil Mungel

The integration of AI agents into enterprise software necessitates new strategies for managing telemetry, token costs, and governance. Agentic workflows significantly increase data volume, requiring robust observability tools to compress and analyze telemetry effectively. Token management has emerged as a critical reliability metric, forcing organizations to implement guardrails, such as AI gateways and automated model routing, to control expenses and prevent runaway agent behavior. A notable example involves a research agent that accrued excessive costs due to an unconstrained objective function, underscoring the importance of human-in-the-loop oversight. Furthermore, standardization through frameworks like the Open Cybersecurity Schema Framework (OCSF) is essential for normalizing high-throughput telemetry data across diverse infrastructure, ensuring that security teams can extract actionable insights from petabytes of information. Nikhil Mungel, head of AI R&D at Cribl, provides these insights on managing the complexities of the current agentic landscape.