ServiceNow Expands AI Control Tower for Agents

🔑 Enhanced Key Takeaways

• •The integration of Veza provides ServiceNow with identity-centric security, allowing the Control Tower to map permissions and access rights across non-ServiceNow AI models and data stores.
• •Traceloop's observability technology enables the Control Tower to perform 'AI tracing,' allowing administrators to debug agent hallucinations and latency issues by visualizing the full execution path of LLM calls.
• •ServiceNow is positioning the Control Tower as a centralized 'AI System of Record' to address enterprise concerns regarding shadow AI and compliance with emerging global AI regulations like the EU AI Act.

🛠️ Technical Deep Dive

• •Utilizes a distributed telemetry collection architecture to ingest logs from external AI agents via standardized OpenTelemetry protocols.
• •Implements a graph-based relationship model to map AI agent dependencies, data access paths, and user permissions, derived from Veza's identity graph technology.
• •Features a 'Guardrail Engine' that applies real-time policy enforcement on LLM prompts and responses, utilizing Traceloop's tracing capabilities to detect anomalies in agent reasoning chains.
• •Supports multi-model observability, allowing for side-by-side performance benchmarking of proprietary models (e.g., ServiceNow Now LLM) and third-party models (e.g., GPT-4, Claude) within the same dashboard.

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline