Dynatrace and Arize AI push observability from detection toward action
SiliconANGLE Sam Weston
Dynatrace bought Arize AI to make observability do more than spot problems. The bet is that AI systems need context, not just alerts, and agents may help fix what they find.
Based on reporting by SiliconANGLE, Sam Weston — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Dynatrace is folding Arize AI into a market that’s moving away from simple alerting. The old model was built for software that behaved predictably, with logs, metrics and traces telling teams whether something was broken. AI systems don’t play by those rules. They can produce different answers from similar inputs, which makes “is it up?” a much smaller question than “did it answer well?”
That’s the gap Arize was built to fill. Its Phoenix open-source platform is used by more than 4,000 enterprises, according to co-founder and chief product officer Aparna Dhinakaran, and Arize AX gives teams a managed place to run AI observability at production scale. Dynatrace brings the broader application side, where agents are calling APIs, touching databases and depending on infrastructure that sits outside the model itself.
That matters because the people chasing these problems are already buried in tools. During the podcast conversation, Practice Lead and Principal Analyst Paul Nashawaty cited research showing that 75% of organizations use between six and 15 observability tools. Add separate AI monitoring and evaluation systems on top of that, and the result can be more sprawl, not less. Dynatrace chief product officer Steve Tack argued the real loss is when teams stop thinking in terms of the whole system.
The more interesting part is where observability is headed next. Dhinakaran said it’s no longer just about humans staring at dashboards, and that these systems are becoming about action. That means telemetry has to be useful not only for engineers, but for AI agents that can diagnose, recommend and potentially remediate problems.
Dynatrace has already been pushing in that direction with Dynatrace Intelligence and its BlueBox AI offering for agentic development and SRE workflows. Arize adds evaluation and AI-specific observability to that stack. Put together, the message is clear: enterprises aren’t just monitoring AI anymore. They’re trying to make it operational.
My take — AI-written commentary, not fact-checked reporting
This is the right direction, and also the dangerous one. Once observability starts feeding agents that can act, bad telemetry stops being a nuisance and becomes a business risk with a dashboard. The industry loves calling this automation; in practice it’s just higher standards with less room for sloppy data.
Read more about this at: SiliconANGLE
