ServiceNow is an American enterprise software company headquartered in Santa Clara, California. The company has more than 29,000 employees worldwide and reported fourth-quarter 2025 subscription revenue of $3.57 billion and fiscal 2026 subscription revenue guidance of $15.53 billion to $15.57 billion.
The company has also invested heavily in AI and automation to improve workflow efficiency and enterprise productivity. ServiceNow acquired Passage AI to strengthen its conversational AI capabilities, expanded its partnership with NVIDIA to support autonomous AI agents, and committed $1 billion through its venture arm to support enterprise software and AI startups.
ServiceNow is also investing C$110 million to support AI adoption in Canada’s public sector, including infrastructure and an AI Center of Excellence.
The company is actively using its proprietary AI platform (Now on Now) to achieve significant ROI and demonstrate how executives can move beyond AI experimentation to tangible, scalable automation.
Here are two key AI use cases powered by Now Assist (GenAI) that enterprise leaders can learn from.
- Reduce agent documentation time with built-in generative AI: Leverage generated AI within your existing IT service management (ITSM) and customer service management (CSM) workflows to automate the creation of summaries, solution notes, and knowledge articles, saving agents time and freeing them to focus on higher-value support work.
- Predict customer escalations: Use machine learning and real-time event monitoring to identify at-risk accounts early, automate proactive responses, and reduce costly customer escalations.
Built-in generative AI reduces agent writing time
A San Jose State University research paper, “Enhancing Customer Service with Generative AI,” reports that customer service agents spend 35% to 45% of their time on repetitive documentation, creating $2.6 billion in labor inefficiencies annually across the United States.
A Harvard Business School study on repetitive task specialization similarly found that service agents lose significant amounts of time documenting and summarizing, drawing focus away from high-value problem solving.
To solve this, ServiceNow launched Now Assist for ITSM and CSM, following the company’s official ITSM Now Assist documentation and CSM Now Assist guide to embed generative AI directly into the agent workspace. Automate incident history summarization, resolution note creation, and knowledge article generation within your existing workflows instead of a standalone chatbot.
The company used machine learning and generative AI to automate routine aspects of support cases, such as summarizing long incident histories, creating resolution notes, and generating knowledge base articles.
ServiceNow Reports: With Now Assist, LLM captures the context of a case, generates editable summaries/notes in seconds, and provides agent reviews in record time, without having to switch context to an external tool.
Chat screen showing a conversation between an IT agent and Now Assist (Source: ServiceNow)
In the same whitepaper, the company shared that Now Assist generates notes within seconds so agents can review and refine them instead of creating them from scratch. This reduced the time required for each resolution note by approximately 80%.
ServiceNow also shared that, on average, ITSM agents save 4 to 6 minutes per use, and CSM agents save 12 to 16 minutes per use. This proves that the value of enterprise AI comes from incorporating generative AI into existing workflows, rather than standalone demos.
Predict customer escalations in advance
Historical monitoring relies heavily on manual checks of tickets and events, making it difficult to spot deteriorating customer experiences before they escalate. Without a scalable way to predict which accounts are at risk, proactive support is inconsistent and often comes too late.
ServiceNow has upended the classic reactive support model by using machine learning to predict and prevent customer escalations. Rather than waiting for a customer complaint or threat of escalation, teams are now leveraging ServiceNow’s unique predictive intelligence and event management capabilities to proactively identify at-risk accounts and respond to issues before they snowball, according to a case study published by ServiceNow.
The effort is built on ServiceNow’s Predictive Intelligence framework, which hosts the underlying machine learning models and event management to capture performance-related events in real-time.
Within Predictive Intelligence, supervised models trained on historical escalation patterns analyze tickets, surveys, CSAT scores, and engagement signals. Event management adds real-time system alerts.
How the workflow works:
- Build and train the model. Past escalation and correlation events are converted into structured features and used to train and validate the XGBoost classifier throughout the PoC/PoV phase.
- Introduce real-time risk scoring. Once the model is up and running, it continuously scores customers as new tickets or events arrive and assigns them an escalation risk label, such as low, medium, or high.
- Automate proactive interventions. When a customer moves into a high-risk category, ServiceNow workflows automatically generate priority alerts, assign follow-up tasks to support or account teams, and display recommended playbooks with next steps.
Screenshot showing the benefits of this solution (Source: ServiceNow)
Over time, the results of each engagement are fed back into the model to continually refine its predictions, whether or not an escalation was avoided.
The business results are also explained as follows:
More timely intervention stabilizes at-risk accounts before they become vocal, resulting in faster response and resolution times, higher customer satisfaction scores, and smoother renewals and upsells.
Before this model, only about 11% of customer engagement was proactive. Since implementation, approximately 68% of engagements have become proactive, allowing for earlier and more structured responses to at-risk customers.
This system has helped us engage with hundreds of customers per year, preventing the majority of escalations while keeping our false positive rate around 3%, and preventing engineering resources from being wasted on unnecessary cases.
