“Without an AI operating model, you cannot survive in this world”: DevOps at Think 2026

At this year’s IBM Think conference held in Boston, DevOps decision makers came together to confront the stark reality of IT in 2026: AI isn’t just one part of their discipline, it is the discipline.

Over the past decade, DevOps has evolved to include ever more granular, probing observability practices. The wealth and volume of data those tools collect requires powerful automated solutions — both for making sense of it all, and for acting on the insights the data provide. 

But while AI-powered observability tools accelerate that process, they also amplify complexity, cost and operational risk.

At Think, IBM’s DevOps leaders introduced new tools for cutting through the morass and coordinating action across increasingly complex environments.

As IBM CEO Arvind Krishna said in Tuesday’s keynote speech, “the models don’t really matter unless the foundation is correct” — and the evolving, AI-powered DevOps pipeline is meant to keep that IT foundation strong. Some key takeaways:

• Hybrid complexity is the default—and AI is accelerating it. An organization’s  environments now span applications, infrastructure, and networks across cloud, on‑prem, and Z systems, and AI workloads intensify scale, cost, and operational risk.
  
  As this complexity becomes the default for organizations, consistent standards and policy‑driven automation are critical for keeping systems reliable. For a deeper look at how teams can build this foundation, see our white paper on standardization and policy‑driven automation.
  
  
• AI is becoming the unifying layer for operations. Rather than creating more silos, AI‑driven operations are emerging as the connective tissue that binds hybrid environments.
  
  
• Observability is moving from visibility to accountability. Teams need real‑time, AI‑assisted insights that don’t just surface issues, but explain impact, causality, and where to act.
  
  
• Application telemetry and infrastructure topology must be unified. Continuously reconciled environments that combine app and infrastructure data give practitioners clearer, end‑to‑end visibility and faster root‑cause analysis.
  
  
• Operational scale now depends on coordinated, AI‑driven automation. As complexity grows, AI‑driven operations are becoming essential to reduce toil, optimize performance and cost, and enable teams to scale reliably.

“One billion new applications in the next five years will come into enterprises because of generative AI,” Dinesh Nirmal, Senior Vice President for Software at IBM, said during a Tuesday keynote address, adding that “Every application that is containerized is going to have hundreds, if not thousands, of microservices” governed by possibly billions of agents.

“How are you going to have a control plane to manage those agents? What is the communication that’s going to happen between those agents, who has access to those agents? What kind of data is being accessed by those agents?”

That same day, IBM automation leaders discussed a new model for observing and controlling agentic IT systems, inspired by one simple truth: The scale and speed of signals have outpaced human response. The IBM Concert platform provides a shared operational layer across applications, infrastructure, networks and security, turning an ocean of data into a continuously updated source of truth for applications, infrastructure, and their relationships.

Concert brings together a suite of IBM capabilities including IBM Instana, Turbonomic and CloudPak for AIOps, among many others, working with — not in place of — existing tools to provide a 360-degree view of the dizzyingly complex AI-powered business environment.

Another DevOps tool unveiled this week promises to make Concert even more powerful, through real-time data visualization: HCP Terraform, powered by Infragraph. 

Infragraph is a real-time database of graphs that give businesses information about infrastructure, applications, data and security in a hybrid cloud environment. When Concert and HCP Terraform powered by Infragraph are used together, the insights gleaned by Concert are grounded in a continually updated resource graph that reflects the current state of infrastructure and the relationships between components.

“Operation at scale, that’s what matters,” Nirmal said while demonstrating the two tools working together onstage Tuesday. “You have visualized it using Infragraph, you have pulled that data into Concert, you can understand it, you can decide, you can act.”

Of course, with greater complexity comes greater demands for security — and the IBM Concert team also announced a new product meant to continuously identify exposure within an agentic IT environment. With IBM Concert Protect and Secure Coder, IBM extends continuous exposure management directly into the developer workflow with a simple loop: detecting exposure across all environments, assessing threats with risk scoring, remediating those threats through automated workflows, and then learning from that data to improve the platform itself.

IBM experts also demonstrated how Instana — the full-stack observability tool powered by agentic AI, now integrated within Concert as Concert Observe — can tackle new challenges of complexity and scale. Jacob Yackenovich, director of product management for Instana, presented Tuesday how IBM Instana delivers automatic discovery of AI components and visibility across hybrid stacks, which helps organizations move from reactive troubleshooting to governed, AI-driven operations.

But while observability is crucial, agentic systems require a broader fabric of monitoring and operations tools to keep track of their complexity in real-time. On Wednesday afternoon, a panel featuring Instana’s Yackenovich with leaders from Marriott and Honda discussed how by pairing the platform with tools like Ansible for provisioning, and Concert for automation of network operations, developers can now create a “unified automation fabric.” 

By constructing a fabric of DevOps tools working together, IT teams can coordinate deterministic, event-driven, and agentic workflows across hybrid environments through a single, policy-driven system, ensuring AI systems operate with the right access at the right time.

In 2026, businesses need DevOps practices that are coordinated, integrated, and comprehensive to keep up with the tidal wave of AI systems and agents swamping the stack. That sense of urgency, above all else, was clear this year at Think, along with a look at some of the IBM tools designed to meet that challenge.

“For every human access in an enterprise, there is going to be 120 non-humans that will be accessing the enterprise,” SVP for software Nirmal said Tuesday.

“How are you going to put privileged access on it? What is your IAM [identity and access management] strategy? How are you going to do policy management? This is why you need an AI operating model … This is the challenge that enterprises are going to see in the next five years.” 

“Without an AI operating model, you cannot survive in this world.”