ManageEngine’s Rakesh Jayaprakash on why the next phase of AI adoption will depend less on experimentation and more on whether organizations have the visibility, security and IT foundations to put intelligence to work
Artificial intelligence has moved from experimentation to executive priority with remarkable speed. Boards are asking where it can create advantage, CEOs are pushing organizations to move faster, and employees who have experienced generative AI personally are arriving at work with entirely new expectations about what enterprise technology should be able to accomplish.
But beneath that enthusiasm sits a more fundamental question: Is the organization's IT environment ready for what it is being asked to do?
For Rakesh Jayaprakash, Director of Product Management at ManageEngine, that question is becoming increasingly important as organizations move from experimenting with AI to integrating it into real business and IT operations.
The challenge he sees is not necessarily the capability of AI itself. It is whether the infrastructure, data, governance and operational maturity underneath it is strong enough to support what organizations increasingly expect AI to do.
"AI adoption has moved or is moving a lot faster than the operational maturity of many of the organizations," Jayaprakash said.
That gap between ambition and operational readiness may become one of the defining enterprise technology challenges of the next several years.
SIGNAL | AI readiness is becoming infrastructure readiness
As AI moves deeper into enterprise operations, the quality of the technology environment underneath it becomes increasingly important.
Organizations that cannot clearly see their assets, dependencies, data, permissions and systems may discover that AI does not eliminate those weaknesses. It exposes them.
The shift is significant because enterprise AI is moving beyond generating information toward interpreting environments, recommending actions and increasingly initiating workflows. That requires context.
An intelligent system needs to understand not simply that a server exists, for example, but what applications depend on it, what business services those applications support and what could happen elsewhere if an automated action changes something.
The emerging signal is bigger than AI adoption itself: the organizations best positioned to capture value from AI may be those that have already done the less glamorous work of creating secure, visible, well governed and interconnected IT environments.
AI cannot understand what the organization does not understand
The attraction of enterprise AI is easy to understand. Consumers have become accustomed to asking systems such as ChatGPT a question and receiving an answer almost instantly. That experience naturally creates expectations that AI should be capable of producing similar results inside an organization.
Jayaprakash cautions that enterprise environments are fundamentally different.
AI operating inside a business needs reliable organizational context. It needs to understand relationships among systems, applications, infrastructure and services while operating within clearly defined security, privacy and governance boundaries.
Without those foundations, an organization may still generate answers and automate processes, but the resulting activity may not translate into meaningful business value.
ManageEngine is seeing organizations that have moved aggressively into AI without achieving the results they expected. In many cases, Jayaprakash argues, the model itself is not necessarily the problem. The operational environment underneath it was simply not mature enough.
His test for organizations considering larger scale AI adoption is striking because it has little to do with the latest AI model.
"If I have to offer one litmus test for organizations to know if they're ready for a larger scale AI adoption, it is to see if they have a well-defined CMDB."
A configuration management database provides an organization with visibility into its technology assets and, critically, the relationships between them.
Consider a server experiencing unusually high CPU usage. An automated system may determine that restarting a process could resolve the immediate problem, but does it understand which applications rely on that server, which business services those applications support and what else could fail as a result?
If those relationships are poorly understood, automation can magnify the problem rather than solve it.
That example points to a larger challenge. IT environments have traditionally grown in functional silos, with network, endpoint, security and service management teams often working through different systems and datasets.
"What most organizations don't have today is a unified visibility across their entire IT infrastructure," Jayaprakash said.
As organizations introduce more automation and intelligence, those divisions become increasingly difficult to sustain. An intelligent system cannot reason effectively about an environment it cannot adequately see.
Start with the business problem, not the AI
There is another mistake Jayaprakash sees organizations making. Leaders begin by asking what AI can do for them, creating an enormous field of possible projects without necessarily establishing which one’s matter.
He recommends reversing the process.
"I would say one big advice to executives is start with the top three problems that you think will make a big difference to your IT operations and your business. Work backwards."
That approach shifts the measure of progress from AI adoption to organizational outcomes.
Instead of asking how much AI has been deployed, leaders can ask which problems became easier to solve, which risks were reduced, which processes became faster and which capabilities now exist that did not exist before.
Not every answer will require artificial intelligence. Jayaprakash points to forecasting as an example where established algorithms may already accomplish what an organization needs efficiently. Applying AI simply because it is available can add complexity and cost without materially improving the outcome.
"AI is not the answer to everything," he said. "But AI can make a lot of things easy and more efficient for you."
For boards and executive teams, that distinction is increasingly important. The strategic question is not whether the organization is using AI. It is whether the organization is applying the right technology to problems that matter.
Innovation cannot come at the expense of security
That becomes even more important when AI begins interacting with sensitive organizational data and operational systems.
Organizations face considerable pressure to innovate while simultaneously confronting growing expectations around cybersecurity, privacy, governance, compliance and resilience.
Jayaprakash describes that balancing act as a difficult tightrope, but his priority is clear.
"If it were me, I would choose security first."
His reasoning is pragmatic. Organizations may receive relatively little public recognition for successfully implementing another AI capability. A significant security incident can create an entirely different level of attention.
Rather than allowing that risk to stop experimentation, Jayaprakash recommends controlled progression. Organizations can begin with smaller projects where data exposure, permissions and security boundaries are understood, establish appropriate guardrails, learn from the results and expand from there.
There is also an architectural reason for proceeding carefully.
Most enterprise systems were designed primarily around human interaction. AI introduces the possibility of machine interactions occurring at far greater speed and scale.
Weak permissions, fragmented data, poorly understood dependencies and loosely governed workflows that may have been manageable when humans performed individual actions can become significantly more consequential when machines begin initiating them automatically.
In that environment, AI readiness and cybersecurity readiness increasingly converge.
Global scale with Canadian relevance
ManageEngine, the enterprise IT management division of Zoho Corporation, has operated for more than two decades and works across areas including endpoint management and security, identity and access management, IT service management, observability and analytics.
That breadth gives the company visibility into many of the systems organizations are now attempting to connect with intelligence and automation.
ManageEngine operates across more than 190 countries, exposing it to organizations at very different levels of technology maturity and to regulatory environments that vary considerably by jurisdiction. Canada and North America are important markets, and the company maintains a local presence while regularly engaging with customers and holding events across Canada.
That combination of global experience and local knowledge becomes particularly relevant when technology decisions intersect with Canadian requirements around security, privacy, governance and compliance.
Despite the breadth of the portfolio, Jayaprakash returns to the same principle he applies to AI: start by understanding the problem.
"Our goal is not to sell all 40 to you, but rather find out what are the main pain points that you're facing today."
The technology comes after the problem has been properly understood.
The organizations that know themselves may have the advantage
Perhaps the most consequential idea emerging from the conversation is that AI readiness may ultimately have less to do with AI than organizations expect.
It is about understanding what technology an organization has, how it connects, who and what can access it, which data can be trusted, which systems support critical business functions and what can safely be automated.
Organizations that have spent years strengthening those foundations may discover that they have already completed some of their most important AI preparation.
Those that have not may find that AI exposes weaknesses that were easier to tolerate when technology environments moved more slowly, and humans remained directly involved in most operational decisions.
"For you to advance or implement more AI functionality, you have to go back to the basics," Jayaprakash said. "You have to ensure that the foundations are in place."
The race to AI will naturally focus attention on models, agents, applications and increasingly powerful forms of automation.
But the advantage may ultimately belong to organizations that can give those technologies something equally valuable: a secure, visible and well understood enterprise in which to operate.
Before asking how intelligent the technology can become, leaders may need to ask a more fundamental question:
How well does our organization understand itself?
Continue the conversation at ITCON Canada 2026
ManageEngine will bring these conversations directly to Canadian IT leaders and practitioners this September at ITCON Canada 2026, with events in Quebec City on September 15 and Vancouver on September 17.
The events bring together CIOs, CTOs, IT managers and technical practitioners to explore the strategic direction of IT operations and security in the AI era, along with product capabilities, roadmaps and practical approaches technology teams can apply within their organizations.
For Jayaprakash, the common thread is the same one running through the broader AI discussion: organizations need secure, resilient and manageable foundations capable of supporting what comes next.