Find your people. Pick a challenge. Ship something real. The CreatorCon Hackathon is coming to the Community Pavilion for one epic night. Every skill level, every role welcome. Join us on May 5th and learn more here.

Now Assist for CSM - Getting Started

FernandoCastro
ServiceNow Employee

on ‎10-04-2024 12:02 PM - edited 2 weeks ago

Now Assist for CSM — Implementation Library Home

Your fast, guided path to
enable Now Assist for CSM
effectively and strategically

This central hub brings together the most relevant guidance across the ServiceNow web universe, from Best Practice Library, ServiceNow University, Product Documentation, Partner Success, Communities to Now Support. It is designed to help you move from activation to measurable outcomes with a clear, intentional approach.

While the interface is intuitive and can be activated quickly, achieving lasting value requires careful detailed configuration, thoughtful sequencing, and knowledge across multiple layers of the platform. This guidance ensures you are not only turning on features but unlocking their full potential in real customer service operations.

 
Product Overview — Understand what Now Assist for CSM delivers across all personas
Icon_CRM_sml.png
Product OverviewUnderstand what Now Assist for CSM delivers across all personas
Customers

Benefit from faster self-service and more relevant answers through conversational interfaces, smart case deflection, and tailored experiences across portals, embedded web, and voice channels. Generative AI powers accurate, contextual responses, while agentic AI orchestrates more complex interactions that span multiple systems or steps.

Live Agents

Work more efficiently with embedded summarization, content generation, contextual recommendations, and sentiment analysis in the CSM Configurable Workspace and Now Assist Panel. Generative AI reduces manual effort for repetitive tasks, while agentic AI plans and executes multi-step actions that accelerate resolution and improve consistency.

Managers

Gain comprehensive visibility into service operations and AI performance through unified dashboards, trend analysis, and outcome tracking. Real-time metrics on trending topics, escalations, and customer sentiment enable data-driven decisions. AI-powered forecasting helps anticipate volume spikes and capacity needs.

Admins

Can deploy, configure, and monitor AI-driven capabilities with minimal overhead, leveraging out-of-the-box skills, low-code customization, and integrations with existing workflows. Generative AI enables fast configuration and knowledge creation, while agentic AI expands automation into orchestration of complex service processes.

This combined approach increases case deflection, reduces handling times, improves service quality, and accelerates time to value.

Take a look at the product details overview here by yours truly:

 
Getting Started with our Implementation Blueprint — Strategic framework for sequencing your Now Assist for CSM activation
ServiceNow_LocationPin_250331.png
Getting Started with our Implementation BlueprintStrategic framework for sequencing your Now Assist for CSM activation

When implementing AI solutions for customer service and your business units, ask the organizations:

1
Do you need to build or buy AI solutions, or both? Which option will give you the best value for your money and the fastest results?
2
What problems are you trying to solve with AI? What are the top use cases? What does your business need, and how can AI improve work for your people?
3
How will you manage the AI you use for your business? Like any technology stack, you want to be consistent in your policies and control them.
Remember: AI is only as intelligent as the data platform it is built on.

The main outcome of this exercise is to define the unique use cases and impact they will have on your business on day 1 and how Now Assist can help bring these to life quickly and efficiently.

For guidance on what to implement first from a best practice perspective, refer to our implementation blueprint. Please note this may vary per customer environment, value prioritization, and business needs. After each stage, update the implementation blueprint to reflect cumulative progress and insights that may pivot the scope.

IBNAFORCSMZP4.jpg

IPLibraryZP4.jpg

 
Training — Build the foundation before you configure
exam.png
TrainingBuild the foundation before you configure
 
Communities — Persona-level articles and readiness tools
26dashboard.png
CommunitiesPersona-level articles and readiness tools

You are right where you need to be. Review each skill article in detail per relevant persona:

Persona Community Article
Customer Self-Service Now Assist for CSM - Customer Self-Service
Agent Productivity Now Assist for CSM - Agent Productivity
Manager Insights Now Assist for CSM - Manager Insights and Visibility
Admin Efficiency Now Assist for CSM - Admin Efficiency
Agentic Workflows Now Assist for CSM - Agentic Workflows
Each article contains definitions, video guides, links, implementation steps, key capabilities, key best practices, how to measure value, and FAQs. Key for generative AI skills is to ground and customize OOTB skills for your use case. See how to here using case summarization as an example.
 

Now Assist Readiness Evaluation App (optional for existing customers)

Download the Now Assist Readiness Evaluation app

 
CRM AI Academy — Monthly live sessions - first Thursday of each month
academy series.png
CRM AI AcademyMonthly live sessions - first Thursday of each month

Join us online every first Thursday of the month at 8 am PT | 11 am ET through our CRM AI Academy Sessions:

Register for Upcoming Sessions

Month Session
September 2025 Picking the Right AI for the Right Task
October 2025 Tailor your AI sequence for Scalable Success
November 2025 Delivering Proven AI Business Value
December 2025 Ensuring AI Success through Evaluation
January 2026 Techniques to ground your AI in CRM — Recording
February 2026 Designing and Deploying Successful Agentic Workflows — Recording
March 2026 Voice Agents - From Intent to Real Time Resolution
April 2026 Introducing AI-Powered Comprehensive Complaint Management in CSM
 
Workshop, Download and Tailor our Best Practice Library Key Artifacts — Best practice assets, templates, and implementation toolkits
assets.png
Workshop, Download and Tailor our Best Practice Library Key ArtifactsBest practice assets, templates, and implementation toolkits

Download, review and tailor our Best Practice Library assets for your specific implementation:

 

Now Assist Skill Kit

Preparation Guide Implementation Guide Scaling Guide
Create your own Center of Excellence and Innovation with training and Organizational Change Management that is AI focused.
 
Now Assist for CSM Capability Matrix — Full catalog of generative and agentic capabilities by category
ServiceNow_AIEL_Icon_Light_250903.png
Now Assist for CSM Capability MatrixFull catalog of generative and agentic capabilities by category

Full generative and agentic capabilities across the Now Assist for CSM capability matrix:

Category Skills and Articles
Customer Self-Service Now Assist in AI Search
Now Assist in Virtual Agent
Now Assist in Engagement Messenger
Now Assist in Portal Case Form
Agent Productivity: Summarization Case Summarization
Chat Summarization and Wrap-Up Generation
Call Summarization
Complaint Case Summarization
Agent Productivity: Recommendations Chat Reply Recommendations
Email Reply Recommendations
Sentiment Analysis
Multi-Turn Q and A in Now Assist Panel
Service Recommendations for Interactions
AI Wrap-Up
Agent Productivity: Content Generation Resolution Notes Generation
Knowledge Generation
Suggested Steps Generation
Activity Response Generation
Manager Insights and Visibility Trending Topics Dashboard
Sentiment Analysis Dashboard
Agentic Workflows and AI Agents Triage Cases Agentic Workflow
Troubleshooting Steps Identification AI Agent
CSM AI Agent: Customer 360 Insights
Complaint Case Intake AI Agent
Accelerate Complaint Case Handling Agentic Workflow
AI Voice Agents for CSM
+ ServiceNow AI Platform agentic workflows
Admin Efficiency View all admin efficiency skills
 

Predictive AI

Category Resource
Manager experience Process Mining for Customer Service Management
 
When to Use Each Type of AI — Match the right AI approach to each service scenario
LLM.png
When to Use Each Type of AIMatch the right AI approach to each service scenario

Select the right approach for each scenario based on complexity, reasoning needs, and automation goals:

Problem Type Recommended Solution Why It Works
Single-step tasks Now Assist Skill Fast, efficient, pre-built
Slightly adapted actions Now Assist Skill Kit Customizable prompts with minimal setup
Multi-step reasoning AI Agent Supports planning, contextual decision-making, and human review
Structured automation Flow Designer Best for predictable, rule-based processes

Read more here

 
Plugin Installation — Prerequisites, plugins, roles, and activation steps
wrench.png
Plugin InstallationPrerequisites, plugins, roles, and activation steps

Before enabling Now Assist for CSM, ensure that your environment and licensing meet the prerequisites and that all required components are properly installed.

1

Verify prerequisites

  • Licensed for CSM Pro+ or Enterprise+
  • Best practice foundational data models or features established (for example, Knowledge, Case Management, Configurable Workspace and Customer Data Foundation or Service Model Foundation)
2

Install required plugins

Navigate to System Definition > Plugins in your ServiceNow instance. Search for and install the following (in order where dependencies apply):

  • Customer Service Management
  • Now Assist for CSM
  • AI Search
  • Optional: Group Action Framework (for grouping of records and taking smart actions in this data)
  • Optional: Knowledge Graph for agentic use in semantic indexing
  • Optional: Generative AI Controller (if using your own LLM models)

Validate that the instance meets version and licensing requirements for each installed feature and you have the latest versions installed per that release.

3

Assign roles and permissions

Before configuring and granting access to skills, make sure you have configured appropriate roles for each user group. Use configurable availability settings to expose skills only where they will add measurable value.

4

Activate through the Now Assist Admin Console

  1. Open Now Assist Admin Console in your instance.
  2. Enable the product plugin from Settings.
  3. Activate the Now Assist Panel for relevant personas.
  4. Configure individual skills within the Customer workflow as required.
Tip: How to Leverage OOTB 'Define Availability' of the Now Assist Skills
5

Customize our reference out-of-the-box Skills and AI Agents

Our intent is to set a standard, guide, and inspire you to ground and evaluate your own skills and AI Agents. Check this video on how to here using case summarization as example.

6

Validate in a sandbox environment

Confirm configuration, data quality, and expected behavior before enabling in production. Test skill execution across intended use cases to ensure both generative and agentic AI perform as expected. We recommend doing this step even when you are trying out the product so you can simulate as close as possible to your actual environment.

 
Security and Responsible AI — Architecture, data handling, and compliance resources (support account required)
security.png
Security and Responsible AIArchitecture, data handling, and compliance resources (support account required)

The following resources require a support account to access.

Overview of Advanced AI and Data Products: Architecture, Applicable Terms, and Data Handling

Start here. This is your foundational resource. It provides a high-level yet comprehensive understanding of the architecture behind our AI products, including Now Assist. You will gain insight into the terms and conditions that apply to their usage, and most importantly, a clear explanation of our overall approach to data handling.

Data Processing for Advanced AI and Data Products FAQ

This FAQ addresses common questions specifically about the processing of data within our Advanced AI and Data Products. It clarifies aspects such as how data is processed and what security controls and processes are in place. This article contains detailed architecture diagrams, data flows, and network infrastructure information.

Data Sharing and AI Model Development

This article clarifies essential questions regarding data handling and AI model development. It outlines how ServiceNow offers optional data sharing programs for model improvement, emphasizing user control through opt-out options and ServiceNow's model development process.

Responsible AI FAQ

This FAQ directly addresses the critical topic of our Responsible AI practices here at ServiceNow. You will find information on the methodologies and principles we employ to avoid and mitigate bias during the training of our Large Language Models and other AI systems.

Addendum to the Consolidated Data Handling and Security FAQs for Generative AI Apps

This addendum provides supplementary information and updates to our existing data handling and security FAQs, specifically focusing on information related to regulated markets and self-hosted customers.

 
Frequently Asked Questions — Licensing, configuration, security, and platform questions
ServiceNow_Networking_250404.png
Frequently Asked QuestionsLicensing, configuration, security, and platform questions
Q. What are the licensing requirements for Now Assist SKUs?
Licensing requirements depend on the existing entitlements a customer has. For instance, a customer must have the Pro or Enterprise base package to purchase the Pro Plus or Enterprise Plus add-on.
Q. What is the difference between Professional Plus and Enterprise Plus SKUs?
They contain the same applications and assist entitlements but are added to different base SKUs.
Q. What is an assist?
An assist is a unit of usage of Generative AI features. Each action performed using Generative AI consumes one or more assists. Assists are tied to the value of an action rather than usage intensity.

Failed requests due to system errors or Guardian blocking are generally not charged, though third-party LLM providers may have their own billing rules. Note that assists are also counted in sub-prod instances.
Find here the number of assists consumed per feature
Q. Do customers need a separate Azure Open AI license to use Now Assist features?
No, the cost is included in the Professional Plus, Enterprise Plus, and Creator Plus SKUs.
Q. Are AI Agents part of Now Assist for CSM?
Yes if they are part of the CSM Agent Collection or part of the Platform bundle.
Q. Can I build my own AI Agents as part of my Now Assist for CSM licensing?
Yes, you can and also mix and match between selected OOTB CSM AI Agents and yours.
Q. Does Now Assist support Regulated Market customers (e.g., FedRAMP, GCC)?
Yes, on a restricted basis with different SKUs compared to commercial environments. For on-premises or isolated deployments, special configuration or approvals may be required.
Q. Which Now Assist skills require a license?
All 'agent' skills require a license, but platform skills for requesters do not.
Q. Can I use a different LLM provider with Now Assist?
Yes. While Now Assist works out of the box with supported model providers, you can also bring your own LLM using the Generative AI Controller. This provides flexibility in model selection based on business, compliance, or performance needs.
Q. How is customer data handled when using Now Assist?
Data used by Now Assist is securely transmitted to processing centers within your region and is not retained. Sensitive data can be redacted before it leaves your instance using the Sensitive Data Handler plugin.
Q. What safeguards can I configure to manage content risk?
Now Assist Guardian allows you to set rules to detect offensive language, sensitive topics, and prompt injection attempts. You can choose to block or allow with logging, and monitor results in the Admin Console.
Q. Does Now Assist support languages other than English?
Yes. Now Assist supports multiple languages natively, and additional ones can be enabled with Dynamic Translation. Language availability may also depend on your chosen LLM provider.
Q. Can I embed Now Assist into custom web or mobile experiences?
Yes. Engagement Messenger supports embedding Now Assist into branded web and mobile environments, with full control over branding, skill access, and fallback routing.
Q. How do I monitor and measure adoption of Now Assist?
Use the Now Assist Analytics dashboard and AI Control Tower to track skill usage, adoption trends, and performance over time. These tools also provide insight into data quality, which impacts AI effectiveness.
Q. Can I restrict certain skills to specific teams or roles?
Yes. Skills can be enabled or disabled for specific roles, channels, or availability windows to ensure they are used only where they deliver measurable value.
Q. How do I ensure changes to Now Assist configurations do not impact production users?
Always validate new skills, role assignments, or configuration changes in a sandbox environment before promoting them to production. This includes testing both generative and agentic AI flows in real scenarios.
Q. Can I enable only certain Now Assist skills instead of all of them?
Yes. Skills can be selectively activated in the Now Assist Admin Console so you can roll out capabilities in phases or target them to specific workflows.
Q. How do I manage updates to Now Assist skills over time?
Revisit skill configuration periodically, especially after platform upgrades, to take advantage of new capabilities or improve prompt tuning and contextual data sources.
Q. Can Now Assist use my organization's existing knowledge content?
Yes. You can connect existing Knowledge Bases, product documentation, or external content sources so that skills can retrieve and generate contextual, relevant outputs.
Q. How should I prepare my data before enabling Now Assist skills?
Ensure knowledge articles, case fields, and product data (CRM Foundation data models) are accurate and up to date. Skills rely on high-quality content to produce relevant and trustworthy outputs.
Q. How do I identify workflows that are best suited for agentic AI?
Look for processes that involve multiple steps, require data from several systems, and vary significantly from case to case. Agentic AI is most effective when tasks cannot be fully scripted with deterministic automation.
Q. Can I combine deterministic workflows with agentic AI in the same process?
Yes. Use agentic AI for adaptive, decision-heavy tasks while keeping predictable, compliance-driven steps in Flow Designer or Orchestration. This hybrid approach balances flexibility with control.
Q. Can you get the Generative AI Controller or AI Agents without Now Assist for CSM?
No. There is no SKU for those features themselves; you need Pro Plus or Enterprise Plus for them.
Q. How are Assists different than Integration Hub transactions?
Transactions are the unit of currency used to measure callouts to 3rd party systems using spokes. Assists are the unit of currency for measuring calls to ServiceNow's private LLMs. If a customer configures connections to public LLM providers (Open AI and Azure Open AI) using the Generative AI Controller, callouts will also consume Assists.
Q. How does ServiceNow protect sensitive data?
Data Privacy for Now Assist is a built-in privacy layer integrated with the Generative AI Controller that masks sensitive data in prompts and responses before data leaves your instance and any GenAI inference occurs. Read more about it here.
Q. Do you need a minimum history case data count to start using Now Assist for CSM?
No, unless you are using GAF-dependent features like Suggested Steps generation, there are no transactional data requirements. Best practices usually lie in the master data architecture.
Q. What transactional data is supported OOTB for generative AI?
Cases and their extensions (case types).
Q. Do you need to migrate transactional data to use Now Assist for CSM?
Depends on the use case but it is not necessary to have transactional data beforehand. You do need to be using CSM to take advantage of Now Assist.
Q. How do we ensure Agent-to-Agent MCP is secure?
  • Authentication and Authorization: Every external agent must authenticate using OAuth 2.0 and adhere to ServiceNow identity controls. Access is governed by roles and scopes, ensuring only authorized agents can invoke specific tools or workflows.
  • Governed Tool Exposure: Our MCP Server acts as a controlled tool plane, allowing us to curate which tools are exposed externally. This prevents unauthorized access and enforces granular permissions for sensitive resources.
  • Data Privacy and Isolation: Customer data remains within their environment. External agents interact through secure APIs without storing or transferring data outside the customer's instance.
  • Monitoring and Logging: All agent invocations, including external ones, are logged in Agent Execution Logs. Dashboards track usage, response times, and failure rates, enabling full auditability and compliance.
  • Protocol-Level Security: MCP and A2A protocols include guardrails for structured interactions, capability discovery, and secure session handling. We continuously update to the latest protocol versions.
  • Governance and Control via AI Control Tower: AI Control Tower provides centralized oversight for agent orchestration, mitigating risks related to cross-platform interoperability and compliance.
Q. What type of governance is in place so people cannot create rogue agents?
We have implemented a multi-layered governance framework to ensure that AI agents are created, deployed, and managed responsibly:
  • Role-Based Permissions and Guardrails: Clear separation of responsibilities between developers, administrators, and compliance officers. Mandatory approval workflows for publishing agents into production environments.
  • Centralized Oversight via AI Control Tower: Visibility into every agent's lifecycle, entity-based permissions, and compliance dashboards to monitor risk and regulatory alignment.
  • Policy Enforcement and Legal Restrictions: Prohibition of unregistered API calls, automated alerts for any deviation from approved governance rules.
  • Compliance and Audit Readiness: Audit trails for every agent action, evidence of data segregation, and alignment with global AI regulations and Responsible AI principles.
 
Past Webinars — On-demand recordings and session archives
microphone.png
Past WebinarsOn-demand recordings and session archives
Join us in person at one of our events such as AI Summits, Knowledge, and World Forums.

Happy implementation!

Comment for any clarifications, additional assets, or assistance needed.

Explore More CSM Articles Join CRM AI Academy
© 2026 ServiceNow, Inc. All rights reserved.
  • 22,312 Views
Version history
Last update:
2 weeks ago
Updated by:
ServiceNow Employee
Contributors