Cisco looses Splunk to probe and tame its growing agentic menagerie

Steven

Cisco looses Splunk to probe and tame its growing agentic menagerie

Summary

Cisco has released an “AI Agent Monitoring” capability for Splunk’s Observability Cloud that visualises agent workflows and monitors the performance, cost, quality and behaviour of LLMs and agentic applications. The new tool will integrate with Cisco’s AI Defense — now generally available — which catalogs Model Context Protocol (MCP) servers, scans for AI‑powered applications, enforces governance guardrails and detects unauthorised LLM programmes.

Cisco also added automated red‑teaming tools designed to run on‑prem so queries and prompts needn’t leave the organisation, and announced a suite of agents for campus, branch and industrial networks (autonomous troubleshooting, continuous optimisation, agentic workflows and proactive zero‑trust recommendations). These capabilities are surfaced in AI Canvas today and will be consolidated into a single Cloud Control management offering later in 2026.

Key Points

  • AI Agent Monitoring for Splunk visualises agent workflows and tracks LLM/agent performance, cost, quality and behaviour.
  • The Splunk capability will be available to users in about two weeks and integrates with Cisco AI Defense.
  • Cisco AI Defense is generally available and can catalog MCP servers, detect unauthorised agentic apps and apply governance guardrails.
  • Cisco added automated red‑teaming tools that can run on‑prem to test models and related infrastructure without sending queries off‑site.
  • Cisco plans agentic features such as autonomous troubleshooting, continuous optimisation and agentic workflows across campus, branch and industrial networks.
  • These features are delivered via AI Canvas now and will be unified into Cloud Control later in 2026; AI Defense is mapped to NIST, OWASP and MITRE frameworks to aid compliance.

Why should I read this

If you run networks or care about AI risk, read this — Cisco’s tying Splunk into its AI Defence stack is a proper move. It gives visibility and control over agentic systems, offers on‑prem red‑teaming so sensitive prompts stay inside your walls, and signals real automation coming to everyday ops. A quick skim will save you time and surprises later.

Context and Relevance

Cisco’s announcements reflect the wider industry push to tame agentic AI with observability, governance and on‑prem tooling. For network engineers, security teams and CISOs this is strategically important: it speeds AIOps adoption, reduces the need to expose sensitive data to external services for testing, and centralises control across formerly separate management planes. The upcoming Cloud Control unification could be a meaningful consolidation for teams juggling Meraki, Catalyst, NX‑OS and ACI management.

Source

Source: https://go.theregister.com/feed/www.theregister.com/2026/02/11/cisco_agentic_ai_update/