Difference between AI agent , Now Assist and Virtual agent.

[sakshinitna]

1. AI Agent:

AI Agents in ServiceNow are like smart digital teammates that understand what needs to be done, make decisions on their own, and take action — all within your workflows. Powered by Generative AI and large language models(LLM), they simplify complex tasks and help you get work done faster, with less effort.

AI Agents are built using:

• Generative AI– for natural language understanding and generation.
• Workflows & Flows – for automating back-end actions.
• Decision Logic – rules, policies, and conditions to act on behalf of users.
• Integrations (via IntegrationHub) – to interact with other systems.
• Now Assist Integration – to interact within Virtual Agent or portal conversations.

How AI Agents Work in ServiceNow

1.You Ask for Help

It starts with a simple request — maybe typed in a portal, spoken on mobile, or asked through a chatbot like Virtual Agent.

"I forgot my password."
"I need to onboard a new hire."
"Show me my pending approvals."

2.It Understands What You Need

The AI Agent listens and thinks:

-“What’s the person trying to do?”

-It uses NLU, Predictive Intelligence, and even Generative AI to figure it out .

3. It Finds the Right Skill

Once it knows what you're asking, it checks its toolkit of “skills” — small, reusable tasks it’s trained to perform.

-Some are flows (like creating an incident)
- Some are scripts
-Some use Generative AI to write responses in a natural tone

It picks the right one for the job, automatically.

4. It Gets to Work

Now it takes action .

- Resets your password
-Retrieves approval records
-Submits your onboarding request
- Or even kicks off a complex workflow behind the scenes

All powered by Flow Designer, scripts, Generative AI, and integrations — seamlessly working together.

5.It Gets Back to You

- “Your password’s been reset.”
- “You have 3 pending approvals.”
- “Your onboarding request is submitted and assigned to HR.

Example Use Cases:

• AI Agents in ServiceNow are used in multiple domains:

Area	Use Case
ITSM	Incident resolution, triage, auto-remediation
CSM	Answering customer queries, creating cases
HR	Leave requests, policy lookup
Developer Tools	Generate flows/scripts from plain English

2.Now Assist:

https://www.servicenow.com/community/developer-forum/now-assist-in-servicenow/m-p/3322580#M1216090

Please refer to this article to know about Now Assist.

3. Virtual agent:

Virtual Agent is ServiceNow’s AI-powered chatbot that handles routine requests conversationally, any time — across portal, mobile, or messaging channels. It uses NLU,Generative AI, and skills to get work done through chat.

Virtual Agent is built using:

• Generative AI + NLU – to understand free-form chat
• Topic Flows & Skills – to automate tasks through guided conversations
• Decision Logic – to make real-time decisions in a chat flow
• IntegrationHub – for system-wide actions (e.g., resetting passwords)
• Now Assist Integration – for contextual Generative AI responses inside the chat

How Virtual Agent Works in ServiceNow

1. You Start a Conversation

On a portal, in mobile, or through Teams/Slack, you just say:

• “I forgot my password.”
• “I want to request leave.”
• “Show my open incidents.”

You type or speak — and the Virtual Agent listens.

2. It Understands the Intent

Virtual Agent analyses your message using Natural Language Understanding:

• Understands intent and extracts relevant details
• Can even ask clarifying questions if required

3. It Selects the Right Topic

It picks a predefined topic or a Generative AI response based on:

• Topic Flows
• GenAI prompts
• Scripting and contextual decision-making

Whether it’s creating a record, providing an update, or answering a question — it chooses the best path.

4. It Performs the Task

Virtual Agent gets to work, like a digital teammate:

• Creates incidents or requests
• Retrieves updates or approval records
• Escalates issues, triggers workflows, or integrates with external tools

All without the user leaving the conversation.

5. It Replies Like a Human

Virtual Agent gives you helpful, friendly responses:

• “Your password reset request is complete.”
• “Your approval is pending with John from Finance.”
• “Leave approved. Enjoy your time off!”

Example Use Cases for Virtual Agent:

Area	Use Case
ITSM	Reset password, request support, incident updates
HR	Check leave balance, submit leave request
CSM	Ask FAQs, submit a customer case

How Virtual Agent, Now Assist & AI Agent Work Together(with Use Case: "Reset My Password)

Step 1: Virtual Agent – Starts the Conversation

The Virtual Agent acts as the front door — it’s how the user initiates the request.

-User Interaction:

"Hi, I forgot my password."

-Virtual Agent Response:

“No problem! Are you trying to reset your password for your work account?”

• It uses Natural Language Understanding (NLU) to detect the intent.
• It may ask follow-up questions like:
• “Is this for your email or system login?”
• “Do you want a reset link or full reset?”

What it does:

• Collects user identity and context (e.g., which account, urgency, contact info).
• Calls the appropriate topic flow or Generative AI response to continue the process.

Step 2: Now Assist – Enhances the Experience with Contextual Help

While the user interacts or waits, Now Assist comes into play by surfacing relevant, in-context support.

Now Assist Suggests:

“Here’s a knowledge article that explains how to reset your password yourself.”
or
“I’ve pre filled the password reset request form for you. Would you like to submit it now?”

What it does:

• Uses Generative AI to summarise and present content like:
• Reset instructions
• Knowledge base articles
• Can autofill forms using record context and user profile.
• Helps reduce steps for the user, especially if manual reset is possible.

Step 3: AI Agent – Executes the Task Intelligently in the Background

Once the user's intent and data are captured, the AI Agent takes over behind the scenes.

AI Agent Executes:

• Validates the user's identity
• Connects with Active Directory or SSO tools
• Triggers password reset
• Updates the user record
• Logs the activity in a task or incident

What it does:

• Uses Decision Logic to decide:
• Is the user eligible for auto-reset?
• Does this need approval or verification?
• Uses IntegrationHub to:
• Talk to systems like Azure AD , LDAP
• Send reset instructions or execute the reset

Once done, the AI Agent reports back to the Virtual Agent:

-“Your password has been reset successfully. You’ll receive a new temporary password via email.”

Who Does What?

Component	Role
Virtual Agent	Interacts with the user, collects info, starts the request
Now Assist	Provides helpful content, form autofill, contextual guidance
AI Agent	Executes the task end-to-end across systems using logic and integrations

Summary

	Ai Agent	Now Assist	Virtual Agent
1. Definition	The doer-understand intent, make smart decisions, and take action across workflows	The helper-Generative AI-powered assistant that suggests or summarises	The talker-Conversational chatbot that guides users
2. Purpose	Executes complex tasks and workflows without help	Helps users complete tasks with smart suggestions	Interacts with users in natural language
3. Tech Behind	Generative AI (reasoning + memory + actions)	Generative AI, LLMs	NLU, decision trees, scripts, Generative AI (now)
4.Best For	Admins, process designers, automation	Agents, analysts, employees	End users needing quick help
5.Example	- Reset password - Onboard employee	- Summarise case - Auto-fill form fields	-Where is my laptop? - Raise incident