Achieve exponential outcomes with 3 vital levels of AI

Levels of AI: woman at a desk doing fist pumps at an open laptop

AI is not a strategy. It's not a new channel, product or goal. It's certainly not better chatbots, meeting summaries or email draft creations.

AI is the most powerful enabling technology to emerge in the past century. It will help people achieve more and will fundamentally change how everything gets done—but only if it's understood and implemented holistically.

AI development has accelerated so quickly that talking about incremental improvements with generative AI (GenAI) is insufficient.

Solutions that incorporate AI as the next iteration of advanced technology, as if it were an accessory, fail to leverage it as the building block for enterprise-wide success that enterprises deserve.

The AI spectrum begins with incremental outcomes and expands to exponential results: Analytical AI, GenAI, Agentic AI

Unleashing the full power of AI

Chatbots and auto-summaries are effective tools that can speed up interactions, increase productivity and reduce costs. Just 10 months ago, we trumpeted these tools as the hallmarks of a successful AI launch. They're important use cases that drive a lot of value. But thinking of AI in this point-solution way puts organisations at risk of missing the opportunity to unleash its full potential.

To realise the full power of AI, think of it as a spectrum from incremental outcomes through exponential results. Along this spectrum are three levels of AI:

Analytical AI is all about machine learning, natural language understanding and tools like virtual agents. Analytical AI can help make predictions and recommendations across the enterprise. Examples include case routing, downsell, propensity to buy and headcount model predictions. While it may be tempting to think we can tick the box on analytical AI and move on to autonomous agents, there's still room to extract additional value from it.
Generative AI changed the game with its ability to create content such as case resolution notes, customer success plans, software code and even protein sequences. GenAI can deliver value to every role in the company by offloading many routine tasks from humans to machines.
Agentic AI is capable of delivering exponential value. While the term "agentic" gets used a lot, there are five key attributes that make Agentic AI different: the abilities to reason, plan, orchestrate, collaborate with other agents and learn. Agentic AI can autonomously and proactively execute end-to-end workflows.

Incremental forms of AI are an important advance in technology—one that you'll see in nearly every piece of software you use. But to reap the exponential rewards AI can offer, companies must adopt all three levels of AI and embed them within a platform that can take action.

AI that makes decisions and suggests things without a very strong integration with execution capability (playbooks, workflows, integrations etc.) will not meet companies' needs.

Exponential AI outcomes require coordination

When implemented at the scope it demands, AI can enable an endless array of independent, intelligent and connected AI agents capable of making decisions and executing material business activities—a state we call multi-agent autonomous AI. And we know it works.

ServiceNow AI Agents are driving an estimated $325 million annualised value across the company, transforming our business by processing 400,000 AI Agent-supported workflows per year. Here are three guidelines to achieving these exponential outcomes:

Take a platform approach that ensures AI is embedded throughout the enterprise. This includes implementing what we call AI Control Tower—a system to manage myriad agents, no matter how simple or complex, in every corner of the business.
Choose one or two enterprise-wide metrics to start measuring the success of AI. At ServiceNow, we track revenue per employee as a proxy for productivity. We also measure speed, or what some call cycle time. The idea is that AI should speed up every workflow in the enterprise: creating code, solving customer issues, onboarding employees, refactoring the supply chain, processing orders…the list goes on. Each department has a handful of speed metrics we're tracking.
Prioritise a unique blend of business acumen and blue-sky thinking in the talent leading this transformation. Every department workflow and persona needs to be reimagined in an AI-first world. It's an unprecedented era in technology, but it's not just technologists who will drive business to the next level. We'll also need creative, forward-looking people with a practical understanding of how business works.

To realise the full power of AI, think of it as a spectrum from incremental outcomes through exponential results. Along this spectrum are three levels of AI.

Making autonomous AI real

I encounter the potential transformative uses of multi-agent autonomous AI every day when I speak to customers. They're not asking me, "How do I use AI?" They're asking questions like:

How can I help my customers resolve their issues more quickly?
How can I help my sales team serve their customers better?
How can I help my employees be more productive and satisfied with their jobs?

There are both incremental and exponential answers to all these questions.

How can I help my customers resolve their issues more quickly?

Incremental: With AI Agents, ServiceNow is achieving 84% customer self-service and threefold improvement in case deflection.

Exponential: By adding autonomous AI Agents, we're automating 37% of our customer support case workflows, and we'll accelerate both the response and resolution times.

How can I help my sales team serve their customers better?

Incremental: AI Agents are helping sellers build successful customer strategies and reduce meeting prep time by up to 42%.

Exponential: Imagine an agent that can analyse all your sales assets and a prospect's information to create a custom pitch deck for each request for proposal (RFP) you receive. ServiceNow is building AI Agents as orchestrators, researchers and content aggregators to proactively recommend cross-sell opportunities and align relevant pitch decks, use cases and approved customer stories.

How can I help my employees be more productive and satisfied with their jobs?

Incremental: At ServiceNow, AI Agents are helping to increase developer productivity by 20%, freeing them to focus on innovation. AI Agents are also improving customer service, HR and IT support agent productivity by 15% per case.

Exponential: We're launching AI Agents that can automatically provide an employee with a series of tasks to initiate. Once the employee approves those tasks, the AI Agent takes action. It can also auto-generate a list of customer questions based on historical case data, integrating analytical, generative, and Agentic AI, and increasing the number of people hours freed, which currently stands at 3 million at ServiceNow.

These use cases are possible on the ServiceNow platform because it's a system of action for the enterprise (literally, as enterprise workflows make up our DNA). AI is at our fingertips—and therefore at your fingertips—to help us accomplish whatever we need to accomplish across the business lifecycle.

How to get started

As our customers evaluate how to use AI to boost their organisations' performance, their employees' work satisfaction and their customers' experiences, I tell them:

Don't get derailed by measurement—focus on value. Did anyone get bogged down by return on investment (ROI) demands when switching their companies from desktops to laptops? How much ongoing measurement was required to prove email was a better option than snail mail? When AI is an enabling technology, like email and laptops, it doesn't require a specific ROI. Individual agents that you build with AI might require it, but trying to measure the impact of AI itself is a fool's errand, like trying to measure the ROI of programming in C++, JavaScript or Ruby. Instead, focus on the value of your AI-enabled productivity boost.
Know your AI. There's a lot of noise in the marketplace, with many service providers touting various GenAI and Agentic AI point solutions. If they're limiting themselves in this way, they're limiting you too. It's crucial that leadership understands the value each level of AI can deliver—and how much more powerful it is when paired with an execution-layer platform.
Stop thinking, "What's my AI strategy?" Instead, think like we do at ServiceNow and consider how AI-powered solutions can drive transformation of business processes, experiences and value realisation across your organisation.

See more results from the implementation of AI Agents on our own platform.