GenAI Observability for Splunk

The GenAI Observability for Splunk app provides enterprise-grade monitoring and analytics for generative AI and large language model (LLM) applications. It collects trace-level telemetry from LLM, RAG (Retrieval-Augmented Generation), agent, and tool workloads, enabling organizations to track performance, costs, errors, and security events across their GenAI infrastructure. The app ingests data via HTTP Event Collector into dedicated indexes and processes trace and metric data through accelerated data models. It delivers observability through specialized dashboards covering operations, cost management, error analysis, LLM performance, RAG analytics, and trace exploration. The app includes preconfigured alerts for cost spikes, error rate thresholds, latency SLO breaches, and token budget monitoring. A Python SDK enables application instrumentation using decorators for LLM calls, RAG pipelines, agent workflows, and tool invocations. The app calculates costs based on token usage and configurable model pricing data, providing budget tracking and optimization recommendations. Security features monitor for PII exposure and prompt injection attempts, maintaining an audit trail for compliance purposes.