AI agents for IT operations are moving from pilot to production. Rajesh Ganesan, CEO of ManageEngine, joins Alan Shimel on TechStrong TV to walk through the new ZIA Agents rollout. Furthermore, he explains why purpose-built small language models are already beating frontier models on real CIO problems.

About Rajesh Ganesan

Rajesh joined Zoho Corp in 1997 as a programmer. Consequently, he has spent nearly three decades inside the same company, first building network management software at Advenet and later launching ManageEngine after the dot-com crash. Today, ManageEngine serves roughly 90,000 customers across six continents and covers service management, observability, identity, endpoint and SIEM.

Why AI agents for IT operations need small models

ManageEngine has been shipping machine learning inside its products since 2012. However, the latest wave is different. Rajesh argues that AI agents for IT operations do not need frontier models. Instead, they need narrow, purpose-built small language models trained for triaging, root cause analysis and posture management.

Moreover, the economics matter. Token-based consumption on frontier models rarely pencils out for high-volume IT workloads. Therefore, ManageEngine hosts and runs its own models as the default across the ZIA family. As a result, customers get conversational, generative and agentic capability without paying a hyperscaler for every alert.

Three use cases inside ZIA Agents

Rajesh walks through three concrete deployments. First, a visual triaging map ingests 20 different alert sources and points teams at the right first fix. Additionally, the agent can execute the remediation autonomously once a human approves. Second, an agent drafts a complete root cause analysis so teams can satisfy regulator requirements without days of manual work.

Third, an agent tackles alert fatigue in the security operations center. Meanwhile, another benchmarks cloud spend against similar companies and flags overpayment. In short, AI agents for IT operations shift teams from firefighting to steering. Read more AI coverage and browse recent TechStrong TV interviews.

The bootstrapped view of AI economics

Zoho and ManageEngine remain bootstrapped, private and profitable. Consequently, the company has a very different lens on AI economics than a VC-backed vendor chasing the next model. That perspective, Rajesh argues, is exactly why AI agents for IT operations belong on a sustainable model layer that the customer can trust for the long haul.