During his recent trip to Dubai, Amit Zavery, President, COO, and CPO of ServiceNow, spoke to us exclusively about how the company is harnessing AI to drive value for its customers.
What brings you to town?
I joined ServiceNow at the end of November—so about four months now. Before that, I spent 24 years at Oracle and the last six years at Google Cloud. I’ve been in enterprise software for a long time, experiencing all the changes in this space. ServiceNow has been amazing so far, with plenty of exciting opportunities over the last four months.
I’m here for customer meetings, including discussions with government entities, who are working on a lot of exciting next-generation innovation projects. ServiceNow has been expanding its presence in this region, and the UAE is a big part of that growth. I’m also meeting with our local team to align on our strategic initiatives in the region.
What are these customers telling you? Are there any common pain points?
Everyone is excited about AI—no doubt about that. But the biggest question customers have is about the outcomes they can achieve.
I met with several customers who are asking how AI can improve their operations, whether it can help predict and prevent issues before they occur, and how it can enhance customer satisfaction by reducing errors and minimizing the need for human intervention.
Ultimately, they are focused on efficiency, automation, and predictability—all aimed at driving better revenue and customer experiences in a cost-effective way.
While everyone talks about the promise of AI, what customers really want to see is measurable, real-world impact.They need tangible use cases and proven results to take back to their boards and say, “We implemented AI, and here are the actual benefits we achieved.”
The key now is delivering AI solutions that provide real, measurable value.
How do you help your customers achieve measurable business outcomes?
ServiceNow stands out because it provides a truly end-to-end platform that integrates seamlessly across an enterprise. Most companies use fragmented systems—separate ERPs, CRMs, and HR platforms—each solving only part of the problem. ServiceNow, however, bridges these silos by acting as a control tower for AI, offering a comprehensive, connected view of business operations.
With this AI-powered control plane, customers gain real-time visibility across their business—both horizontally (east to west) and vertically (north to south). ServiceNow integrates with leading data warehouses such as Snowflake, Databricks, Google BigQuery, AWS, and Microsoft Azure, as well as enterprise applications like Salesforce, Workday, and Oracle. Operating across AWS, GCP, and Azure, our modern, unified platform eliminates the need for patchwork solutions, ensuring one seamless data model that enhances efficiency, automation, and decision-making.
By enabling dynamic, workflow-driven business processes, ServiceNow provides real-time insights and actionable improvements. This is why most Fortune 5000 companies rely on ServiceNow—not just for IT, but also for CIOs, CISOs, security officers, and CFOs who need centralized visibility and control. Our dashboards help organizations track key metrics, drive strategic decisions, and extract real business value.
Looking at the current AI landscape, there’s a gap between AI environments, data, and language. How do you bridge that gap?
Right now, there are two key areas we’ve focused on in AI, as it requires multiple components to function effectively. While large language models are essential, AI also needs to understand intent, orchestrate user requests, apply reasoning, and then break down those requests into AI-driven tasks that interact with backend systems. Managing this entire workflow is a critical part of our approach.
To support this, we have developed Workflow Data Fabric—a system that includes both an internal database and connectivity to multiple data warehouses and enterprise applications. This integration ensures that the data fed into our AI engine is comprehensive and well-structured.
Additionally, with 20 years of workflow data, we have a deep understanding of expected outcomes, allowing us to train AI models effectively. We leverage prompt engineering to refine responses and seamlessly integrate AI into our platform—not as a standalone feature, but as a core component. This means that anything built on top of our platform automatically benefits from these AI innovations and investments.
Can you explain what Workflow Data Fabric is? Is it similar to a data streaming platform, or does it function differently?
It’s a combination of both. First, it serves as an integration platform, using what we call a zero-copy data architecture—meaning you don’t have to physically move data. Instead, we extract metadata and insights from different systems to provide a comprehensive, end-to-end view of your information.
Second, we have an Integration Hub, which includes connectors to various applications and systems. We call these connectors “spokes,” as they enable seamless communication between different platforms.
Finally, we also focus on orchestrating data movement, ensuring that information flows efficiently across systems without unnecessary duplication or complexity.
So, think of the Data Fabric as the engine that connects all the different data sources. The challenge today is that many organizations either can’t collect data efficiently, struggle to interpret it, or don’t have real-time, complete insights. As a result, they can’t take meaningful action based on that data.
Our approach is to bring all three components together—data collection, interpretation, and actionability—into a unified platform. Often, when people receive data, they simply send an email, which isn’t truly useful. The key question is: Can you act on it? That’s why we are integrating an action framework directly into the Data Fabric platform, ensuring that insights drive real, measurable outcomes.
What do you mean by AI agents? Is that different from agentic AI, or are they the same thing?
There are multiple aspects to it. Agentic AI is the overarching concept—it consists of several components. First, it’s a system that understands user intent and breaks it down into actionable steps. For example, if an employee says, “I’m going on maternity leave, what do I need to do?” this request involves multiple tasks, such as updating the HR system, adjusting payroll, modifying benefits, handling role transitions, and delegating pending tasks. In a traditional setup, the employee would have to manually enter information across various systems, or a human agent would have to manage these interactions manually.
What agentic AI does is take that user intent, break it down into multiple tasks, and then create AI agents to handle each one. For instance, it would assign AI agents to update payroll, adjust benefits, and make necessary changes in employee performance management systems. These AI agents complete specific tasks, while an orchestration engine consolidates everything and delivers an outcome. It doesn’t just provide information—it executes the tasks and confirms that everything is completed.
The true power of agentic AI lies in its ability to reason, understand, and orchestrate complex workflows. It spawns AI agents, manages them, and ensures that tasks are completed end to end. In some cases, if human intervention is required, it can even incorporate human agents into the workflow for supervision. The key value of agentic AI is in automation, predictability, and completeness—reducing the burden on employees and ensuring seamless task execution.
