ServiceNow Prepares for the Company of 2033: Autonomous Agents, AI Control Tower, and Processes to Tame AI Chaos

"We are in the era of AI chaos," introduces Bill McDermoth, CEO and Chairman of ServiceNow, at Knowledge 26, the annual event where the company unveils updates related to its platform. The multinational's mission is to bring order to this chaos, ensuring visibility, observability, and control over AI. Relying on this technology is now impossible, but continuing to govern AI, models, and agents with traditional tools is also unthinkable. For this reason, ServiceNow has enhanced the AI Control Tower by adding new connectors, integrations, and tools aimed at better managing corporate AI systems, making them more secure, and, most importantly, autonomous.

This does not mean that ServiceNow is becoming an AI company: on the contrary, the focus remains firmly on processes. AI, in this case, becomes just a new process, or rather a system of processes, to be orchestrated more effectively.

The Future is Made of AI Agents: It’s Time to Learn to Govern Them

AI will be increasingly essential, and the future will consist of artificial intelligence agents that will not just suggest but make decisions on behalf of the users themselves. McDermoth envisions the future of automation, which is not merely the classic visionary conception from tech gurus. According to the CEO of ServiceNow, we are rapidly approaching an inflection point: declining birth rates in the Western world will make it impossible to further increase the global workforce, and the only way to sustain growth and innovation will be to rely on intelligent machines. McDermoth does not speak of a distant future but points to a specific date for this inflection point, namely when the number of agents and robots will surpass that of human workers: the year 2033. It is practically just around the corner.

Based on these considerations, McDermoth is shaping ServiceNow's automation platform to adapt to this new world. However, it’s important to note that ServiceNow does not intend to become an AI company. It does not aim to compete with those developing AI models or solutions. The enhancements to the AI Control Tower are precisely in this light, helping companies first discover which AIs and agents are running in order to evaluate their behavior, potential security risks, and economic returns.

This isn't new: solutions addressing this have been available for quite some time. However, ServiceNow's ambition goes beyond, aiming to surpass the "sidecar" approach, where AI acts as a co-pilot, a virtual employee offering suggestions that humans then implement, moving towards enabling AI to execute actions independently. Supervised, of course, but totally independent. The key terms are three: autonomous workforce, autonomous CRM, and autonomous analytics.

To realize this vision, the company is already working on a series of agents that will be available by 2026, specializing in specific business functions: finance, IT, cybersecurity, HR, legal, procurement. However, the journey is just beginning, as McDermoth emphasized in his keynote address: "Only six out of ten companies have started using AI agents. And of those, only one in ten has achieved tangible results."

All Companies Use AI, but Few Projects Surpass the Pilot Phase: What Future for AI?

We have repeatedly reported on Edge9 studies showing that, at present, few companies have been able to fully harness the advantages of AI. While many have understood the potential of Large Language Models (LLMs), using them daily for personal productivity, when it comes to large-scale implementations of AI agents fully integrated with business processes, the situation changes. There is a huge obstacle that few companies have managed to overcome. In fact, there is more than one obstacle. The first lies with data, which is often scattered across various systems and poorly organized, making it difficult for AIs to work. But that is not the only issue. The main problem lies in the processes.

Filippo Giannelli, Country Manager for Italy at ServiceNow, explains this with a car analogy: "You can have a very powerful engine, but you need a chassis behind it, because if the engine is fantastic but the chassis can't support it, everything stagnates." The chassis, in the case of AI, is a workflow capable of channeling information properly, enabling productivity to increase safely. And to build this chassis, much work must be done on business processes. "AI is a powerful automation tool when it is well channeled. If it isn’t, it can be a tool of chaos and cost explosion," the manager emphasizes.

Hence, the need to establish well-defined processes where AI can be channeled to limit errors, misinterpretations, and off-topic responses. And to minimize mistakes, especially in crucial stages. "A 3% error rate can be acceptable in some internal processes," Giannelli explains, "but in other areas, it can be a disaster. Consider a navigation system that indicates the wrong route 3% of the time: this is obviously unacceptable." The goal of AI Control Tower is precisely to help understand how to reduce errors and to keep track of AI expenditures, which can now easily exceed those for the employee’s labor costs, and especially to ensure regulatory compliance, a topic that is increasingly hot.

And contrary to what some may think, humans are not at risk, according to the Country Manager: "AI can help identify inefficiencies, but humans are needed to find business opportunities. Without the human ability to pick up on the non-verbal language of opportunity, technology alone is insufficient." The AI Control Tower aids precisely in this phase: to verify that checklists are followed and respected by machines and that processes are executed correctly. However, to set them up, optimize them, and uncover business opportunities, the human brain remains irreplaceable due to its creativity and lateral thinking capability.

Not Just AI: The Role of Traditional Automation

Some entrepreneurs dream of seeing companies entirely led by AI. An example is Jack Dorsey, founder of Twitter and now at the helm of Block. The company recently cut 4,000 employees (40% of the workforce) because, according to Dorsey, AI tools allow for more effective work with leaner teams. However, this is not a vision shared by ServiceNow, which has a different goal in mind: "We believe we can free people for higher-productivity activities. Often people end up acting as a sort of middleware, mediating between isolated systems and reading long emails or log files. By removing this repetitive and draining work, people can focus on what they are passionate about," explains Terence Chesire, VP of Customer & Industry Workflows at ServiceNow.

Chesire also emphasizes a very important concept: AI is a tool, not an end. Working in the tech sector, we often see how AI is presented as a solution to everything. However, there are cases where traditional automation processes can be more effective and cost-efficient: "We sent our teams to work directly with clients and realized that many of them are still learning where to apply it. I often tell clients: 'This doesn't need to be an AI process.' If a process is standard and deterministic, a classic workflow is faster and less expensive. AI should be reserved for the more important tasks, like understanding unstructured data. We are not reducing engineers but accelerating roadmaps to develop more software. The key is the combination of traditional automation and AI. Using only one or the other would be incorrect."

He adds: "We are focused on results, not on maximizing token usage." This approach contrasts with some other industry giants that suggest evaluating engineers’ performance based on the number of tokens they use.

Chesire agrees on the importance of defining processes before implementing any technology and identifying the performance indicators to monitor: given that "before even AI, you need an end-to-end vision and meaningful KPIs." In this regard, ServiceNow not only positions itself as a technology provider but also as a consultant. Even through its partners, this helps introduce the company’s technologies into SMEs, which often lack the necessary expertise and technological resources to manage complex platforms like that of ServiceNow. As Giannelli explains, "For SMEs, our strategy is based on a partner ecosystem that provides pre-packaged and processed solutions to ensure they integrate into the process." And it assists them in reorganizing and structuring their processes because "the lack of information is 90 out of 100 times a lack of a structured workflow."

The Future is Played Out on the Balance Between Humans and Technology

At Knowledge 26, ServiceNow did not showcase any groundbreaking new technologies, but rather incremental innovations to its platform, focusing on taking small steps to realize a broader vision. The path is defined, aimed at increasingly autonomous operations, with a focus on enhancing the platform to support companies in their AI adoption journey. Helping them scale proof of concepts rapidly and transform them into company-wide projects. In short, it is creating the conditions for AI to truly enter companies, not just in the form of virtual assistants but as agents capable of taking complete control over certain processes. However, to achieve tangible results, companies must invest in improving internal processes, also considering that AI is not always the answer to all problems: in many cases, traditional robotic process automation remains unbeatable.

The vision proposed by ServiceNow is certainly interesting, but the ambitions to put people at the center, as positive as they are, risk clashing with the reality of a market that inevitably looks at cold numbers. And in this historical moment, the risk is that technology will prevail, erasing the human element instead of enhancing it. It should also be noted that if by 2033 the inflection point has indeed been reached, agents will become an objective necessity.