We've updated the ServiceNow Community Code of Conduct, adding guidelines around AI usage, professionalism, and content violations. Read more

Gen AI Skills, AI Agents & Agentic Workflows for FSO

mmcdonnell
ServiceNow Employee

2 hours ago

Gen AI Skills, AI Agents & Agentic Workflows for FSO

 

A note before you begin: FSO is built on top of CSM, so customers automatically inherit all Now Assist for CSM capabilities. This article focuses on FSO-specific considerations — for broader implementation guidance, skill and agent details, and platform best practices, refer to the Now Assist for CSM – Getting Started article. 

 

Objective

This article explains what Gen AI Skills, agentic workflows, and AI agents are in the context of ServiceNow, and introduces the out-of-the-box capabilities built specifically for Financial Services Operations (FSO).

 

What Is a Gen AI Skill?

A Gen AI Skill is a purpose-built, prompt-driven capability that uses generative AI to assist agents directly within their workspace. Skills are designed for focused, single-step tasks—surfacing a summary, drafting a response, or generating a recommendation—without requiring autonomous multi-step reasoning. Think of them as AI-powered assists: they reduce the cognitive load on agents by delivering the right information or output at the right moment, keeping humans firmly in control throughout.

 

What Is an AI Agent?

An AI Agent is an autonomous capability that can reason, take actions, and produce outcomes on behalf of a user for a specific task or decision point. Unlike a Gen AI Skill, an AI Agent doesn't just surface information—it actively evaluates context and executes actions. Individual AI Agents operate independently, focused on a single domain or step: investigating a merchant, checking regulatory guidelines, or drafting a customer communication. They are the building blocks of more complex automation.

 

What Is an Agentic Workflow?

An agentic workflow is what you get when multiple AI Agents are coordinated by an orchestrator to complete a broader, multi-step process end-to-end. The orchestrator determines which agents to invoke, in what sequence, and how to pass context between them—enabling the platform to handle complex decisions and actions autonomously while maintaining governance and compliance. The key distinction: if there's an orchestrator directing a team of agents, it's an agentic workflow. If a single agent is acting on its own, it's an individual AI Agent

 

Now Assist Gen AI Skills for FSO

FSO ships with three out-of-the-box Gen AI Skills designed to reduce manual effort and help agents and adjusters act faster with better context.

 

Dispute Case Summarization

Managing high volumes of dispute cases means agents often spend significant time reviewing case history before they can act. The Dispute Case Summarization skill uses generative AI to condense a dispute record into a concise, high-level summary—giving agents the pivotal context they need to understand a case at a glance. The skill uses a tailored prompt designed to cover a wide array of dispute categories and is within the agent workspace. This helps financial institutions minimize ambiguity, reduce time-to-action, and improve customer satisfaction through faster, more informed resolution.

 

To see this in action, view our feature short on dispute summarization.

To activate: Now Assist for FSO product documentation

 

Claims Case Summarization

Insurance claims can be complex, involving multiple lines of business, detailed documentation, and layered policy information. The Claims Case Summarization skill condenses a claim record into a brief, high-level summary to help claims adjusters quickly find the information they need. The skill uses tailored inputs and prompts that address specific lines of business and is accessible from the Now Assist panel—keeping adjusters in their workflow while AI does the heavy lifting of synthesizing case context.

To see this in action, view our feature short on claims summarization.

To activate: Now Assist for FSO product documentation

 

Disputes Intake via Virtual Agent

One of the most significant friction points in dispute management is the intake process itself—gathering the right information from customers in a compliant, efficient way. The Disputes Intake via Virtual Agent capability enables cardholders to report a dispute through a conversational, AI-powered dialogue rather than waiting on hold or navigating complex forms.

 

The Virtual Agent asks contextually appropriate questions—such as which transaction is being disputed and the reason for the dispute—and can rephrase questions conversationally if a customer's response is unclear. Behind the scenes, it navigates card network rules and issuer policies to determine exactly what information is required. It can also pull in transaction data from core banking systems, accept supporting files (photos, emails), and flag cases as ineligible for deflection based on network rules.

This capability was built with banking-specific compliance requirements in mind:

  • Now Assist is grounded in pre-defined case intake requirements and cannot go off-topic with customers.
  • Full visibility and control over the model is maintained by the institution.
  • A conversation log is stored on the case for proper auditability.
  • The Virtual Agent collects required customer consents before proceeding.

The result is a complete, ready-to-resolve dispute case populated without agent involvement in intake—reducing contact center volume and accelerating time to resolution.

To activate: Now Assist for FSO product documentation

 

FSO Agentic Workflows

Agentic workflows go further than skills—they can autonomously execute multi-step processes, make decisions, and initiate actions, with humans staying in control via guardrails and oversight.

 

Resolving Friendly Fraud Disputes

FSO includes an out-of-the-box agentic workflow designed specifically to help resolve friendly fraud disputes. This workflow:

  • Automatically gathers relevant transaction and customer data.
  • Uses predefined decision rules to validate the dispute.
  • Supports compliance with Visa Compelling Evidence 3.0 (CE 3.0) requirements by gathering and validating the qualified transaction data needed to contest first-party misuse—including prior eligible transactions, device/IP data, and matching payment credentials—enabling dispute liability to shift from the issuer to the merchant when criteria are met.
  • Initiates actions such as refund processing or case escalation when required.
  • Provides notifications and reporting to keep stakeholders informed throughout the process.

By automating these steps, the workflow reduces manual effort, accelerates resolution times, and ensures consistency across disputes.

To see this in action, view our Feature Short on Resolving Friendly Fraud Disputes here.

To activate: Now Assist for FSO product documentation

 

Resolving ACH Disputes

FSO powers ACH dispute resolution with a coordinated team of four AI agents working under an orchestrator. While built and configured for ACH, the underlying agent architecture is extensible—making it a strong foundation for other payment types and dispute categories such as wire transfers, P2P payments, or check disputes. Organizations can adapt the agent logic, knowledge sources, and decision criteria to fit their specific payment rails and operational requirements.

The ACH agent team consists of:

  • Merchant Analysis for Disputes AI Agent — Investigates the merchant associated with the disputed transaction and produces an outcome on merchant credibility.
  • Nacha Operating Guidelines Check AI Agent — Performs the checks prescribed by Nacha guidelines for the disputed transaction, referencing a knowledge base article to determine eligibility and compliance.
  • ACH Dispute Return Recommendation AI Agent — Drawing on the merchant analysis, Nacha guidelines check, and similar past cases, recommends a course of action for the dispute outcome.
  • Dispute Communication AI Agent — Based on the selected outcome, communicates with the customer or the merchant's bank and creates a task to track the action.

Together, these agents automate the end-to-end investigation and communication process. The workflow automatically gathers relevant ACH transaction, authorization, and customer account data; applies predefined decision rules and regulatory criteria to validate the dispute; initiates actions such as provisional credit issuance, return processing, or case escalation; and provides notifications, audit trails, and reporting throughout the process. By automating these steps, the workflow reduces manual effort and ensures consistent, compliant handling of ACH cases—while serving as a reusable template for other payment dispute workflows your institution needs to support.

To see this in action, view our Feature Short on ACH Disputes with Agentic AI.

To activate: Now Assist for FSO product documentation

 

How AI Agents Power Agentic Workflows in FSO

Agentic workflows are powered by multiple AI-driven capabilities that work together to gather context, analyze data, and drive actions. Some agents focus on collecting transaction and customer information, others classify disputes and surface recommendations, and others help initiate actions like credits, returns, or case escalation. Investigators stay in control, with AI providing guidance, supporting evidence, and reporting to ensure fast, consistent, and compliant dispute resolution.

 

Creating Your Own Agents in ServiceNow

ServiceNow allows organizations to build custom AI agents to support their unique workflows. Using you can define the tasks your agent should perform, the data it can access, and the actions it can take. Custom agents can gather and analyze context, provide recommendations, automate routine tasks, or generate notifications—all while keeping humans in the loop for oversight. By creating your own agents, you can extend ServiceNow's out-of-the-box capabilities to meet specific operational needs, streamline processes, and ensure consistent, compliant outcomes.

 

Whats Next:

For further context and hands-on learning, explore the following:

 

 

0 Helpfuls
  • 18 Views
Version history
Last update:
2 hours ago
Updated by:
ServiceNow Employee
Contributors