Beyond automation: Dynatrace unveils agentic AI that fixes problems on its own

AI unsurprisingly took center stage during the keynotes and throughout much of Dynatrace Perform 2026, the annual user conference last week in Vegas. But while AI now plays a profound role in DevOps, platform engineering, and IT in general, its role in observability is rapidly becoming critical for those who monitor systems and take action based on AI-augmented observability data.

In Dynatrace’s case, with the release of Dynatrace Intelligence, the company takes a bold step toward offering agentic AI as part of a core operations infrastructure. Dynatrace uses the word “autonomous,” although it goes well beyond that.

Its agentic AI platform for operations not only detects problems and interacts to help resolve them — with human oversight, of course — but also extends toward a future vision often discussed: a world in which agentic AI systems interact directly with other AI agents for observability.

That vision is no longer theoretical. It is already beginning to manifest with this week’s release of Dynatrace Intelligence. As Dynatrace describes it, Dynatrace Intelligence is an agentic operations system at the core of the platform and serves as the reasoning and decision-making layer for all the agentic layers built upon it.

Dynatrace Intelligence fuses deterministic AI with contextual analytics to ground agentic decisions in real‑time facts and to create more reliable agentic workflows. It also coordinates different kinds of agents while minimizing hallucinations so organizations can trust automated actions.

During the keynote, Dynatrace CEO Rick McConnell outlined what he described as “From reactive to autonomous” for his vision of “AI-powered observability.” He said this is mandatory to navigate this complexity. He outlines the evolution of IT operations through three phases:

• Reactive Operations: The traditional model of fixing incidents after they break.
• Proactive Operations: Using automated root cause analysis to reduce Mean Time To Repair (MTTR).
• Preventive Operations: The current “phase three” utilizes predictive AI to anticipate and resolve issues before they impact users.

Big complexity

Needless to say, scaling adds complexity, and that complexity demands observability.

In numbers, the rapid growth of hyperscalers like AWS, GCP, and Azure now account for annualized revenue exceeding $286 billion, McConnell said.

This growth has led to an explosion of data and a radical increase in complexity, resulting in tool fragmentation and making it harder for teams to manage IT environments with fewer resources.

To solve this, the industry must evolve from reactive operations to preventive operations, and finally to autonomous operations. This requires an integrated platform that delivers deterministic answers rather than guesses, enabling trustworthy automated actions, McConnell said.

“It’s not about AI replacing you. It’s about AI making every one of us more effective.”

“If software has to work perfectly, what that means is that it couldn’t break in the first place, and the only way that we can handle that is through an AI-integrated platform,” McConnell said. “We want to find the low-hanging fruit to deliver autonomous, trustworthy, and then we want to expand from there to make it better and better.”

From cloud monitoring to cloud operations

The transition from simple monitoring to active cloud operations is a given. While organizations have vast amounts of data, they often lack the ability to take action on it without human intervention, Steve Tack, Dynatrace’s chief product officer, said during his keynote.

The objective is to move away from just generating alerts and toward “agentic remediation,” where data and answers drive automated responses, Tack said.

“One of the common themes that we hear is that we’re drowning in data, but we’re still starving for action,” Tack said. “We’re really driving this to make it a simpler experience, to onboard unified visibility across all the different telemetry types, so that we can shift you into proactive operations.”

Enterprises want AI that can act, “not just advise,” Tack tells The New Stack via email. “For teams to move confidently from supervised runbooks to trustworthy, self‑healing operations governed by SLOs and policy, they need to be able to ground automation in deterministic, causal insight and a live understanding of dependencies and impact,” Tack says.

What’s next

While aggregate market spend was approximately $189 billion for AI just a few years ago, projections for the next seven years indicate an explosion to nearly $5 trillion, McConnell said during his keynote. This shift is framed not as a threat to human employment, but as a necessary evolution to handle increasing complexity, McConnell said.

He emphasized that the goal is to enhance human capability rather than replace it, noting that organizations are seeking ways to use AI to amplify their teams and capabilities.

“It’s not about AI replacing you. It’s about AI making every one of us more effective, more productive in our work,” McConnell said. “We have to deliver a reliable AI because we have to be certain that the output, the results, are meaningful, and that they are headed in the right direction.”

The post Beyond automation: Dynatrace unveils agentic AI that fixes problems on its own appeared first on The New Stack.