AI
AI Is Scaling Faster Than Enterprises Can Monitor It

AI is generating more telemetry, adding new operational complexity and pushing observability budgets higher. In the UAE, only 32% of respondents say observability has helped them resolve issues faster, while Saudi organizations report stronger AI deployment and security outcomes.
[For more news, click here]
AI can make mistakes; it can behave unpredictably, consume more resources than needed, or take actions that don't align with policies. These are important factors to monitor and correct as more AI agents go into production.
New Dynatrace State of Observability 2026 research finds that AI adoption is increasing both the volume of operational data enterprises must manage and the expectations placed on observability platforms. In the UAE, 46% of respondents say AI has increased telemetry volumes, while the figure reaches 43% in Saudi Arabia, compared with 33% globally.
The numbers suggest that AI is creating a new operational problem alongside the opportunity it promises: organizations are deploying more intelligent systems, but understanding what those systems are actually doing is becoming harder.
AI Is Changing What Enterprises Need to Monitor
Globally, Dynatrace found that 99% of respondents use AI in their operations, while almost half now run between six and 25 production AI projects, up from 42% a year earlier.
57% now rank AI-powered capabilities such as anomaly detection, autonomous operations and root-cause analysis as important when choosing an observability platform, while 49% specifically prioritize visibility into AI workloads and LLM behavior.
Regarding investments, Dynatrace found that 78% expect observability budgets to increase over the next year. Among those organizations, 96% say AI adoption is at least a moderate driver of that increase. Nearly half already spend $1 million or more annually on observability.
AI adds another layer of complexity. Across the global survey, 43% identified the expansion of AI, LLMs, and agentic workloads as a leading source of technology complexity, ahead of multi-cloud environments and security threats. Only 6% reported no significant visibility gaps, while AI infrastructure, agent activity, interactions between AI and conventional systems, model behavior, and resource costs dominated the areas where organizations struggled to see clearly.
The UAE Has a Visibility Problem
The issue is particularly apparent in the UAE. Almost half of UAE respondents (49%) say unpredictable AI application and model behavior has increased complexity. Another 45% cite managing multi-cloud and hybrid-cloud environments.
Forty percent cite the cost of storing, processing and retaining observability data as a challenge, while the same share struggle with integrating information from multiple sources. But the more important number is the outcome. Only 32% of UAE respondents say observability has helped them detect and resolve issues faster, compared with 52% globally. Just 33% say it has helped accelerate AI projects from pilot into production.
This suggests organizations are gathering more operational data without consistently using it to make decisions faster.
Security is another gap: 38% of UAE security respondents miss their own targets for fully remediating critical vulnerabilities, the highest share among the 12 markets surveyed. Nearly a third take between one and four weeks on average to fully fix a critical vulnerability.
The challenge, then, isn’t just collecting more telemetry. It’s making that data useful enough to explain what happened, how AI contributed to the problem, and what teams should do next.
Saudi Arabia Follows a Different Pattern
Saudi organizations report a somewhat different experience. Only 25% say AI has increased the overall complexity of their technology environment, roughly half the global figure. However, 43% say AI has increased the number of monitoring tools they need.
37% of Saudi respondents cite managing sensitive or regulated data inside observability pipelines as a challenge, above the global figure of 31%. That matters as Saudi organizations deploy AI into more regulated environments, where telemetry itself can contain sensitive information about applications, users and business processes.
48% say observability has helped move AI projects from pilot to production, slightly above the global figure, while another 48% report better-informed business decisions.
Security remediation is also faster: 35% of Saudi security respondents say they fully remediate critical vulnerabilities within 24 hours, compared with 28% globally.
The comparison with the UAE is notable. In both markets, AI is increasing telemetry volumes, but the operational pressure differs. UAE respondents report greater difficulty translating observability into faster issue resolution, while Saudi respondents place more emphasis on governing sensitive information as they scale AI.
Enterprises Want Autonomy Before Full Visibility
Organizations are rapidly moving toward autonomous operations. 87% already use AI agents moderately or extensively for observability and IT operations, while 44% say they want fully autonomous operations within five years. Only 12% report being at that level today.
But enterprises still rely heavily on human verification. Respondents say humans verify an average 66% of AI-powered decisions, while only 27% currently address visibility into how AI agents make decisions and act. Across nine AI observability capabilities, current execution trails perceived importance by an average of 14 percentage points.
That leaves an uncomfortable gap. Organizations want AI agents to take on more operational responsibility while many still lack the evidence needed to explain what those agents did and why.
Dynatrace found that businesses increasingly consider real-time AI output quality, attack detection, and model-drift monitoring important, but implementation continues to lag behind expectations. Skills shortages and fragmented tooling remain the most commonly cited barriers.
They also argue that enterprises now need both observability of AI, which verifies behavior, quality, security, cost and impact, and AI-powered observability, where AI itself helps analyze and operate the wider technology environment.
For the UAE and Saudi Arabia it is apparent that AI adoption itself is not the difficult part, it is whether the organizations have built enough monitoring around these systems to idenfiy when they are working and where the humans need to intervene. As agentic AI takes on more responsibility, that evidence may ultimately determine how much autonomy enterprises are prepared to give it.
Related Articles
Dynatrace Looks to Close the AI Observability Gap as Enterprise Adoption Grows
Why AI Is Creating a Data Crisis Inside the Enterprise
AI Is Turning Data Sovereignty Into an Enterprise Infrastructure Problem





