Cisco’s 2026 State of Observability report finds that AI adoption is outpacing enterprises’ ability to monitor agent quality, control unpredictable costs and connect AI investments to business outcomes.
Enterprises are accelerating the deployment of AI agents, but their ability to monitor performance, manage costs and maintain trust is struggling to keep pace. According to the Splunk State of Observability 2026 report released by Cisco, 87% of respondents globally consider monitoring AI a priority, yet 56% say they cannot effectively measure critical AI workload metrics such as large language model response quality.
Based on a survey of 2,050 IT operations and engineering professionals across 10 countries, the report highlights the growing importance of agent observability as organisations move AI agents into software development, customer experience and operational workflows.
The gap is particularly relevant in India, where 85% of respondents identify AI monitoring as a priority. However, 39% lack clear business outcomes tied to these efforts, while 46% say they cannot measure AI workload metrics such as response quality. Although 87% of Indian respondents consider visibility into AI hallucinations important, only 70% say they are effective at monitoring them.
“Managing agents at scale requires reimagining both security and observability.” — Jeetu Patel, President and Chief Product Officer, Cisco
Unpredictable AI Bills Complicate Business Cases
AI economics is emerging as a parallel challenge. Globally, 55% of respondents say they cannot control AI costs, while 60% report that unexpected billing spikes create more operational strain than high baseline costs. In India, the corresponding figures are 43% and 54%.
Telemetry itself is adding to the pressure: 57% of respondents say its costs have exceeded the cost of the infrastructure being monitored. As enterprises run more agents across multiple models, tracking consumption and attributing expenditure to specific workloads becomes increasingly important.
Kamal Hathi, SVP and GM, Splunk at Cisco, said observability can help organisations turn AI from an unpredictable expense into a measurable, strategic investment by introducing guardrails and incorporating monitoring into AI token economics.
Security and Observability Converge
The report also highlights closer collaboration between observability and security teams as AI agents gain access to more enterprise systems. Globally, 81% say monitoring AI workloads has increased this collaboration, rising to 89% in India. Meanwhile, 73% of respondents globally say they can accurately trace performance issues to security root causes, up from 65% in the previous year.
The stronges