Why Context Matters More Than Automation in IT Operations: A Q&A with Shriram Iyengar

AI is everywhere in IT operations right now, but results are mixed. Some teams are seeing real improvements in uptime and incident response. Others are adding more tools without much to show for it. In this Q&A, Shriram Iyengar, Vice President of Business Development, Sales & Leadership, Global Growth, shares what’s working, where things tend to break down and what leading organizations are doing differently. From operational debt to self-healing workflows, he explains why the future of IT operations isn’t just about more automation.

A: The best support system is one that never needs to be called. If everything is working the way it should, users don’t even think about IT operations. In fact, one of our clients described it really well. They said, with us, they can “celebrate silence.” That idea stuck with me. When everything is running smoothly, there’s nothing to escalate, nothing to chase. That silence is actually a sign that the system is working exactly as it should.

But behind the scenes, there’s a lot going on. Teams and systems are constantly monitoring application health, infrastructure, integration and data flows. They’re looking for early signals—anything that suggests degradation—so they can fix before they cascade. I often compare it to a duck swimming. On the surface, everything looks calm. Underneath, there’s constant movement to keep things stable. When that coordination breaks down, issues don’t stay isolated. They cascade across systems, and that’s where the business impact shows up—very quickly.

A: Most organizations have implemented some form of automation—scripts, runbooks, Robotic Process Automation (RPA). These usually solved very specific, point problems. The real issue is that real-world problems don’t happen in isolation. They cut across systems, teams and workflows. So, you end up with automation at different steps—alerting, ticket creation, maybe even resolution—but nothing connecting those steps end to end. And as systems and business processes change, those automations quickly become outdated. We’ve seen cases where teams had to maintain entire functions just to keep their automations relevant. Automation helps at a micro level. But without coordination across the full lifecycle, it doesn’t deliver meaningful impact.

A: AI is most useful when you’re dealing with scale and complexity, especially the volume of data coming from monitoring systems. There’s a constant stream of signals: alerts, logs, events. AI helps reduce that noise, correlate what matters and accelerate diagnosis. From there, it can identify patterns and even recommend or trigger preventative actions. Where it falls short is when it’s applied in silos. If you’re using AI on just one dataset or one part of the incident lifecycle, you’ll get limited results. The real value comes from connecting data across systems—applications, infrastructure, tickets, changes—and understanding how they relate. That shared context is what allows AI to move from analysis to action.

A: Operational debt builds up when new releases and changes keep adding work for support teams, but the underlying issues never get fully resolved. Every new feature introduces new dependencies, new failure points, new activities. Over time, teams are dealing with both the new work and all the old, repetitive issues that were never fixed at the root. It compounds quickly. We’ve seen situations where support teams are effectively months behind—handling current releases while still dealing with problems from much earlier ones. At that point, you either keep adding people, or it starts impacting the business. Reducing operational debt is critical because it’s what frees teams from constantly reacting and allows them to actually improve the system.

A: Because systems are no longer simple, and they’re constantly changing. An issue is rarely just one thing. It’s usually the result of multiple systems interacting—applications, infrastructure, integrations—all influencing each other. To diagnose that, you need to understand how everything is connected. That’s where most organizations struggle. If you don’t have a clear view of dependencies and service relationships, you’re essentially troubleshooting symptoms instead of identifying root cause. Once you bring that context together—through service mapping and a connected view of the ecosystem—you can get to probable root causes much faster.

A: Technology isn’t the hardest part. Most organizations know what tools they need. The bigger challenges are around people, process and change. There’s often resistance to changing how teams work. Over time, organizations develop habits and workarounds that aren’t always efficient, but they’re familiar. There’s also a knowledge problem. A lot of operational knowledge sits with individuals—people who know where to look, where to check, how to fix things. When you introduce a platform that makes that knowledge more accessible and standardized, it changes how teams operate. That takes adjustment. If adoption doesn’t happen—if people fall back to old ways of working—then even the best technology won’t deliver results.

A: At a basic level, self-healing means the system can detect an issue, apply a known fix and validate recovery without manual intervention. But it’s not a one-time setup. A common misconception is that once you build a self-healing workflow, it will just keep working forever. In reality, systems change, so those workflows need to evolve as well. The better way to think about it is this: Automate what’s repeatable, involve humans where judgment is needed. With agent-based approaches, you can have systems recommend actions, execute in known scenarios and bring in humans for the exceptions. That balance is what makes self-healing effective.

A: We’re moving from AI-assisted operations to more autonomous models. Earlier, AI was helping with specific tasks—analysis, recommendations. Now, we’re seeing the emergence of agent-based systems that can work together to solve problems end to end. This idea of agentic “swarming”—multiple agents collaborating based on context—is a big shift. It’s what enables the transition from sensing to detecting issues to deciding and acting on them in a coordinated way. That’s what will drive more autonomous operations at scale.

A: Traditional metrics like response time and resolution time are still important, but they’re not enough. You need to look at outcomes. For example: Are you reducing repeat incidents? Are issues getting resolved the same day? How much work is being handled by AI versus humans? How often do you still need human intervention? You also need visibility across the full stack—from infrastructure to business processes—so you can see where bottlenecks occur and how they impact the business. Ultimately, the goal is to reduce the cost to serve and shift effort from “run the business” to “change the business.” That’s where real impact shows up.

As enterprise environments become more connected—and more complex—IT operations can no longer rely on isolated tools or reactive workflows. The organizations seeing the most progress are shifting toward AI-driven operating models that connect signals, context and action across the full operations lifecycle.

That shift isn’t just about faster incident response. It’s about reducing operational debt, preventing repeat failures and creating systems that continuously improve over time. Platforms like Sapient Sustain are designed to support that transition by combining enterprise context graphs, agent-driven workflows and human oversight into a more connected approach to IT operations.

Why Context Matters in AI-Driven IT Operations | Publicis Sapient

Why Context Matters More Than Automation in IT Operations: A Q&A with Shriram Iyengar

Meet the author