Splunk survey finds AI agent monitoring lags deployment urgency
Cisco's sixth Splunk observability survey of 2,050 IT and engineering respondents finds agent oversight, AI cost control and security collaboration gaps.
The story in focus
Cisco released the Splunk State of Observability 2026 report, surveying 2,050 IT operations and engineering professionals across 10 countries. It found that urgency to deploy AI agents outpaces monitoring ability: 87% call AI monitoring a priority, but 54% lack clear business outcomes tied to it and 55% say they cannot control AI costs.
- 87% of respondents say monitoring AI is a priority, but 54% report no clear business outcomes tied to it.
- 55% say they cannot control AI costs; 60% say unexpected billing spikes strain operations more than high baseline costs.
- 81% say monitoring AI workloads increased observability-security collaboration, and 84% now share data between those teams, up from 74%.
Monitoring gap widens as agents spread
Cisco published the sixth edition of its Splunk observability research, based on an Oxford Economics survey of 2,050 IT operations and engineering professionals across 10 countries, conducted from February through March 2026. The report argues that adoption of agentic AI is outpacing organizations' ability to watch over it.
Respondents report uneven oversight. While 87% call monitoring AI a priority, 54% say they have no clear business outcomes tied to it, and 56% say they cannot measure AI workload metrics such as LLM response quality. On hallucinations, 85% say visibility matters, but only 58% call themselves effective at it. Cisco says 83% are concerned about unintended consequences of AI agents.
Costs and security collaboration
Financial predictability is a central theme. Fifty-five percent say they cannot control AI costs, and 60% say surprise billing spikes cause more operational strain than high baseline charges. Fifty-seven percent report telemetry costs exceeding the infrastructure being monitored; 67% use a single observability platform.
The report also tracks convergence between observability and security teams. Eighty-one percent say monitoring AI workloads increased collaboration, and 73% can trace performance issues to security root causes, up from 65% a year earlier. Troubleshooting with security teams rose to 77% from 62%, and data sharing climbed to 84% from 74%. Cisco's Jeetu Patel said managing agents at scale requires rethinking both security and observability.
Leader practices
Cisco separates a leader group it says achieves concrete outcomes by using AI agents for frequent remediation, monitoring model quality, cost, usage, security and privacy, and tying telemetry to business impact. It reports leaders are likelier to attribute AI costs accurately, 91% versus 69%, and 54% say observability has yielded at least $5 million in annual business revenue, against 28% of non-leaders.
What share of surveyed organizations say they cannot control AI costs?
Fifty-five percent of the 2,050 ITOps and engineering respondents say they cannot control AI costs, as AI use grows more complex with rising consumption and hard-to-predict usage, according to the Splunk report.
Sources & further reading
- Splunk State of Observability 2026: Hidden Costs of AI Agents ↗newsroom.cisco.com
Help keep the record accurate. Send a correction.
