Ronak Desai, Founder and CEO of Ciroos – Interview Series

Ronak Desai, founder and CEO of Ciroos, leads the company with a clear mission to eliminate IT toil and give time back to SRE, DevOps, and operations engineers. He brings a deep conviction that AI should meaningfully augment human expertise rather than replace it, especially in high-stakes operational environments. Before founding Ciroos, Desai spent more than 20 years at Cisco, where he held multiple senior leadership roles, including Senior Vice President and General Manager of Cisco Full-Stack Observability and AppDynamics. Across his career, he has focused on building scalable, customer-centric platforms, holds more than 50 patents in active use today, and carries forward the principles of innovation and customer obsession that shaped his tenure at Cisco.

Ciroos is building an AI-native SRE teammate designed to dramatically reduce the time it takes to investigate and resolve complex IT incidents across modern, multi-domain environments. The platform uses native multi-agentic AI to reason across signals, automate investigations, and support automation, augmentation, and autonomous operations—while ensuring humans remain firmly in control. By correlating data across tools and domains that are traditionally siloed, Ciroos enables teams to move from reactive firefighting to faster, more confident decision-making, freeing engineers to focus on higher-impact work rather than repetitive and exhausting operational toil.

You spent more than two decades at Cisco, helping to build some of its most successful networking and observability products. What inspired you to take the leap and found Ciroos?

Throughout my interactions with various enterprise teams, I saw the same storyline play out repeatedly. Operations teams were overwhelmed by dashboards, chasing alerts and depending on institutional knowledge to troubleshoot issues across multiple systems. Despite significant capital being spent on observability, they still lacked a way to connect evidence across domains in real time. My co-founders and I wanted to change that. We set out to build an AI system that could reason like an experienced operator and work in concert with SREs from the outset, enabling teams to focus on improving resilience and reliability rather than spending time searching for insights or fire-fighting issues.

You’ve described Ciroos as a response to one of the toughest problems in operations — investigations that span multiple domains. How did your experience leading the AppDynamics and Full-Stack Observability business at Cisco shape that realization and influence the architecture of Ciroos?

At AppDynamics, we achieved a high level of insight into application behavior. However, when the cause of an incident lived outside the application (whether in cloud configuration, networking or IAM), having visibility at the application layer alone was insufficient. The challenge was in establishing context. That experience guided how we designed Ciroos. Our platform brings AI reasoning to scale production operations. It looks at signals across domains, aligns events on a common timeline and reasons across domain boundaries to determine the true root causes of incidents.

Ciroos introduces the concept of an “AI SRE Teammate.” How does this idea of AI as a collaborator differ from traditional automation or observability tools?

The AI SRE Teammate functions more like a new teammate than a new tool. It listens first, gains an understanding of the environment, accepts defined assignments and fosters trust over time. While traditional automation executes rules, the teammate applies reasoning. When it identifies an issue, it selects the relevant domain expert agents to query, gathers supporting evidence and presents it in context. This collaborative element frees up engineers’ time to validate and problem-solve rather than manually derive correlations.

Your platform uses multi-agentic AI reasoning. Can you explain how multiple AI agents coordinate to accelerate root cause analysis and improve accuracy across complex systems?

Each agent has domain expertise — one in Kubernetes, another in cloud, another in networking and so forth. When an incident occurs, these agents work together as part of a central reasoning layer that correlates findings in real time. The system determines which agents to invoke, what tasks to assign to each agent, in what order, and for how long. This coordination reduces investigation times and improves accuracy by ensuring that every layer is evaluated in context rather than in a silo.

From a technical perspective, how does Ciroos dynamically reason across disparate data sources — such as cloud telemetry, application logs and infrastructure metrics — without overwhelming users with noise?

Ciroos considers every data source as a single lens in a larger picture. It aligns observations across data sources on a unified timeline and surfaces only the relevant causal relationships. For example, if a pod restart event occurs after a small change in IAM or network policy, Ciroos automatically connects that sequence. It goes beyond providing raw dashboards and instead assembles a complete story based on the evidence that helps engineers understand why something happened.

Trust and explainability are central to your design philosophy. How do you ensure that AI-driven recommendations remain transparent and that human engineers stay firmly in control?

Each recommendation comes with the supporting evidence and the reasoning that led to it. Engineers can trace each conclusion, test their assumptions and manage the system’s level of autonomy, from assistive to semi-autonomous. The system retains contextual knowledge over time through human feedback, allowing it to improve decision quality while remaining fully governed. Our approach resembles the way a team would onboard new teammates, with clear guardrails, direct reasoning and full human oversight. Trust builds as the system shows increasingly reliable performance over time.

Early adopters report that Ciroos reduces investigation time from hours to minutes. What kinds of patterns or insights surprised you most when teams began using the AI SRE Teammate in production?

There have been two pleasant surprises — first, the speed at which even large enterprises have responded favorably to our core value proposition has been heartening. Second, our customers have looked closely at our technology and have come up with some very unique use cases that go well beyond root cause analysis. These use cases highlight the real-world challenges that large enterprises face today in their production operations.

The term “AI as a Teammate” suggests collaboration rather than replacement. How do you see this concept evolving as organizations grow more comfortable working alongside intelligent systems?

We view this as a journey involving automation, augmentation and, ultimately, autopilot. Although Ciroos supports all three modes today, we typically see organizational adoption of AI following a maturity curve. To begin, enterprises use our AI system to automate clearly defined and repeatable tasks while minimizing cognitive overload for humans. In contrast, bespoke non-AI native systems put too much burden on the human operator to configure lots of parameters and rules before customers realize value.

In the next phase, enterprises leverage the AI system to augment a human’s reasoning at scale across multiple domains, even as the system provides detailed explanations and recommendations for remediation that the human validates and executes. This is where most enterprises are today.

Over time, the AI can manage full incident workflows autonomously for the enterprise, only escalating to a human when necessary. We expect this to be gradually opened up based on the task. That progression is similar to how teams develop trust with new hires. As you gain more confidence, the partnership grows deeper.

Many enterprises already rely on established observability and incident management platforms. How does Ciroos integrate with these existing ecosystems without disrupting workflows?

From the beginning, integration was never going to be optional. We believe that a federated data model provides enterprises the fastest time-to-value, most optionality and lowest total cost of ownership. The Ciroos AI SRE Teammate integrates with seven different categories of enterprise systems today — observability, incident response, collaboration tools, cloud platforms, ticketing systems, CI/CD tools and physical infrastructure via open APIs and protocols such as MCP and A2A. It integrates into established workflows instead of requiring teams to adopt new ones. This design has helped make it easy for enterprises to adopt. Teams get quicker answers without changing their existing workflows.

You’ve emphasized customer obsession and innovation throughout your career. How do those values guide Ciroos’s culture and its long-term vision for redefining reliability engineering?

Being customer-obsessed means being relentlessly focused on the real-world challenges faced by our customers’ operations teams, such as long hours, fatigue, toil and the constant search for answers to questions that come up in operations. Innovation is about solving those problems in ways that meaningfully return time and focus. We envision all operations teams having an AI teammate that learns continuously, scales with demand and helps ensure reliability across systems. In the long term, we see AI service as software becoming standard across the entire development to production operations cycle — systems that think, act and improve alongside their human peers. If we can provide our users with the clarity and breathing room they’ve always needed, we’ve done our job right. These users could be SREs, IT Operations staff, production operations engineers, cloud operations engineers or DevOps team members performing production operations.