Dynatrace has acquired Arize AI, bringing AI observability, evaluation, and agent monitoring into its application observability platform. The deal, announced September 11, 2026, signals a shift in how enterprises monitor software: applications are becoming less predictable, and observability tools are moving from telling humans what broke to helping AI agents fix it.
Traditional observability was built around deterministic software - logs, metrics, and traces from systems that behave the same way every time. AI applications do not. Large language models and autonomous agents can produce different outputs from similar inputs, which means operations teams need to evaluate response quality, not just uptime and latency.
"Evaluating no longer just becomes about is it right or wrong," said Aparna Dhinakaran, co-founder and chief product officer of Arize AI, in an interview with theCUBE Research's AppDevANGLE podcast. "It becomes about actually measuring the quality of the responses, which is just a very fundamentally different problem."
Why AI and application telemetry need to merge
AI agents rarely run in isolation. They call APIs, query databases, and depend on cloud infrastructure. When an agent misbehaves, the root cause often lives in the traditional software stack around it - not in the model itself.
Dhinakaran said Arize customers wanted stronger connections between AI telemetry and production application telemetry. Dynatrace customers were asking for deeper AI observability. Merging the two gives developers, SREs, and AI engineers a shared view across the entire stack.
"The agent systems and the software systems are joined at the hip," Dhinakaran said. "Having this ability to not only debug agents with AI observability, but also have all the context of the software that they use to call tools or the underlying infra behind the agents … just makes us build better products."
This consolidation also addresses tool sprawl. Nashawaty cited research showing 75% of organizations use between six and 15 observability tools. Adding separate AI monitoring systems risks creating another isolated operational layer. Steve Tack, chief product officer of Dynatrace, said combining the two domains gives teams "a system mindset" instead of forcing them to piece together information across disconnected platforms.
Observability shifts from dashboards to action
The bigger change is who consumes observability data. For years, engineers looked at dashboards and responded to alerts manually. AI agents create a different model: telemetry becomes context that software agents read directly to diagnose problems, recommend changes, or start remediation.
"Observability is no longer about humans looking at dashboards and metrics and logs," Dhinakaran said. "It's about action."
That raises the stakes for data accuracy. Autonomous operations only work if organizations trust the telemetry feeding those decisions. Tack said precise analytics and trustworthy answers will be essential as enterprises give agents more operational responsibility.
Dynatrace has been moving this direction through its AI and automation strategy, including Dynatrace Intelligence and BlueBox AI for agentic development and SRE workflows. Arize adds evaluation and observability for the AI systems participating in those workflows. Arize's open-source Phoenix platform is used by more than 4,000 enterprises, and its managed Arize AX environment targets teams running AI in production. For teams focused on AI for Operations, the acquisition points to a future where monitoring tools feed both human engineers and the agents increasingly helping operate systems.
Why this matters for operations teams
Operations professionals should expect observability platforms to become machine-consumable infrastructure, not just human-facing dashboards. The practical implication: telemetry you collect today may soon be read by agents that recommend or execute fixes automatically. That means data quality, context, and evaluation become operational requirements, not nice-to-haves. Teams that treat observability as a shared system of record - spanning AI behavior and traditional infrastructure - will be better positioned as AI moves from experimentation into production. Professionals in AI for IT & Development roles will need to understand both sides of that stack.
Your membership also unlocks:
