Bringing autonomy to enterprise IT

Praneeth Vanteru, Technical Evangelist at ManageEngine, spoke to CXO DX about the company’s growing focus on agentic AI, its use cases across IT management and security, and its approach to data sovereignty, governance, and automation.

How is the growing focus on agentic AI changing ManageEngine’s approach to AI capabilities across its product portfolio?

Traditionally, organisations, and not just ManageEngine, have treated AI as a set of features. Now, the focus is slowly shifting towards business outcomes. It is about what organisations can achieve using these AI capabilities. So, as a product company, we are moving from providing a set of features to providing certain outcomes, which is what businesses are looking for.

With Zia agents being introduced across your product suite, which of the various IT operations are becoming autonomous, and what level of autonomy can customers expect?

I can give you a few examples. In a ticketing system, we have introduced an agent for L1 tickets. Most of the issues that are repetitive can be handled by the L1 agent on behalf of the technician. Another example is creating knowledge articles. Most technicians are so tied up with working on tickets that they don’t have enough time to create knowledge articles. That is something an agent can handle.

If you look at security, there are so many logs coming in that it becomes difficult to carry out threat analysis manually. An AI agent can handle detection and investigation. In Log360, we now have SOAR capabilities, so the agent can also act on a threat. In observability, we have something called causal intelligence, where we don’t just detect that something is down. We can also identify the root cause of why it is down, so that you address the root cause rather than just the symptoms.

These are some of the areas where we see AI evolving. We are also building more capabilities around identity and endpoint management. For example, if we detect certain threat activity on an endpoint, an agent can act on it, quarantine that particular endpoint, and raise a ticket.

You mentioned identity as another area of focus. What developments can we expect there?

Identity is something we are looking into very deeply because, with all the features and everything AI can do, the biggest risk is how to control it or restrict it to its own role. We have recently released Identity360, which is a cloud-based solution that provides a unified perspective on identity, where you are taking care of different personas, including the agent.

ManageEngine has developed its own Zia LLM and AI infrastructure rather than depending on external models. What advantages does this offer customers?

At ManageEngine, we are a very privacy-focused company. If you look at how we have operated from the beginning, we don’t acquire companies. We build everything step by step. In AI, what we felt was that customers should not be put in a position where they have to send their data somewhere else to benefit from AI features. What we are doing is bringing AI within their tools. First, we can guarantee that we will not use their data to train our models. That is in our hands. Second, they don’t have to send their data anywhere. Since we have built the LLM and the agents internally, we have greater control over the guardrails and the sovereignty aspect, which is very helpful for customers.

Much of ManageEngine’s development takes place in India, but you have customers across different countries. How does your approach to data sovereignty work for customers in markets such as the UAE?

We have data centres across the world. As we speak, we have over 20 data centres globally. In the UAE, we have data centres in Dubai and Abu Dhabi. Our offices are in more than 30 countries, and we operate in more than 190 countries. We also work through our partner ecosystem.

Coming to data sovereignty, take the UAE as an example. Since we have data centres in Abu Dhabi and Dubai, the LLM and AI features are also sitting in UAE data centres. The data doesn’t have to leave the UAE. So, customers in government and critical sectors can benefit from AI while we have full control over the technology, and their data is not going out of the region. That is what I mean by sovereignty. We can guarantee what we have developed while also complying with local regulations.

With Agent Studio allowing enterprises to build their own agents, how do you prevent that flexibility from creating another layer of ungoverned AI?

The good news is that organisations are now realising that AI is just like a human. It needs to be guarded. Everything has to be observed and monitored.

Within our identity solution, what AI is doing and the guardrails around it are part of the identity solution. That is the first thing. Second, since whatever we provide is developed in-house, we have our own guardrails that we have built. On top of that, there is the data sovereignty aspect that we are bringing in. With all these things coming together, we are confident that customers will get the required features without creating another layer of complexity on top of what already exists in their organisations.

ManageEngine talks about data governance, identity and access management as important pillars. How do these translate into practical guardrails around autonomous actions?

When we say data governance, it simply means what data an AI agent can see. Once you have restrictions on what it can see, the next thing is what it can act on. What actions can the agent perform? That depends on the role of that particular agent. The third thing is that when an agent performs an action, there has to be an observability element where you know everything that is being done. So, when we talk about all these terms, what we are saying is that we know what data it can see, we know what it is allowed to do, and we know what it is currently doing. When you have control over these three things, you can get the full value of AI without risking your data or any other component of your network.

How far has AI progressed from assisting analysts to independently responding to threats?

If you look at SOC teams, there are too many alerts coming in. As a human, it takes a lot of time, especially to correlate those alerts. Let’s say there is a ransomware attack on a particular laptop. The attacker will not directly come and start attacking the laptop. He would have tried accessing the firewall, then tried to spread within the network, and eventually managed to encrypt the system. This entire correlation can be done directly through an AI agent, which saves a lot of time.

In Log360, since we have now introduced SOAR, it also gives us the capability to act on those threats. We have an agent for investigation, and we have also introduced playbooks. For different types of threats, the playbook knows what action it has to take within the environment. Combining these two gives you SOAR capabilities with limited human intervention.

For example, if there are particular tasks that could affect production or cannot be reversed, you can have human approval in between. These capabilities are currently available to existing customers.

ManageEngine has a broad portfolio across IT management and security. How are you bringing these products together to provide a unified experience for customers?

We are enabling the orchestration of workflows across different products. Currently, our products are standalone, but they have native integrations with each other, so workflows can be achieved across different products.

For example, let’s say HR creates a new employee. That employee needs to be added to Active Directory, given an email account, added to Microsoft Teams and provided access to certain applications. There are also requirements from other departments. Logistics needs to provide a place to work, and HR needs to create payroll records.

When HR creates a new employee in its system, it can trigger a ticket in ServiceDesk Plus. ServiceDesk Plus can then create different tickets for different departments.

Within IT, because we provide several solutions, it can integrate with Active Directory, create the user, assign an email address based on certain criteria, add the employee to a Teams channel and provide application access depending on the team they belong to. All of this can already be automated through native integrations.

What we are working towards is a single screen where these solutions come together through orchestration. For example, you could define what automation you need in privileged access management and what automation you need in Endpoint Central, and build that orchestration within a single console. That is not available yet, but we are working on it.

How do you assess the current level of agentic AI adoption among enterprises in the UAE? Is it moving into actual deployment, or is it still largely a conversation?

What I have seen is that, in technology, conversations happen first, and then the technology follows maybe six months or a year later. We start talking about something, and then it begins coming into the enterprise. As of now, I don’t think there are enough agentic AI deployments. But if you compare the situation with last year, organisations are now very serious about implementing these technologies and reducing their workloads.

What were ManageEngine’s key priorities at Ai Everything Abu Dhabi as an exhibitor?

We showcased our new enhancements in agentic AI, especially across service management, security, endpoint management, and observability. We wanted to show how we are bringing agentic AI into these areas and how it can help organisations improve their efficiency. At the same time, we wanted to highlight that we have local data centres, so it is not just about the AI features. We are also addressing the data sovereignty aspect. That is the conversation we are looking to drive in the market.