Justin_Wilburn
ServiceNow Employee

We're excited to announce that AI Voice Agents are now live on the Now Platform!

 

AI Voice Agents bring natural, human-like conversations to your customer and employee experiences—answering questions, resolving issues, and completing tasks through voice interactions. Whether it's resetting a password, checking incident status, or managing travel expenses, AI Voice Agents transform each call into an intelligent and effortless experience by understanding context, adapting to intent, and delivering results without friction.

JustinWilburn1_1-1765486722726.png

 

Built specifically for the Now Platform, AI Voice Agents integrate seamlessly with your existing systems to automate service requests, reduce call volumes, and deliver quick, personalized support at scale. With customizable voice options, multilingual support to serve customers across the globe, and comprehensive analytics that track key metrics and sentiment from each call - you have everything you need to create exceptional voice experiences with seamless live agent transfer when needed.

 

 

We've compiled answers to the most frequently asked questions to help you get started

 
*Full FAQ Attached*

 

GENERAL

 

Q: What are the requirements for using AI Voice Agents?

A: To utilize AI Voice Agents on the platform, ensure you have the following:

  1. Now Assist Entitlements
  2. Required apps/plugins installed
    • For Platform level AI voice agents, you will need 'Now Assist for Platform
      • Note: This is Auto-Installed from any of your 'Now Assist for [x Workflow]' apps such as 'Now Assist for CSM' or 'Now Assist for SPM'
    • To utilize OOTB AI Voice Agents, you will need workflow specific plugins
      • IT Service Management AI voice agent collection (sn_itsm_voice_aia)
      • HR AI Voice Agents (sn_hr_voice_aia)
  3. A supported Contact Center Integration with systems such as Genesys Cloud or Twilio. Additional details can be found here

 

Q: What Voice Agents are included Out of the Box (OOTB)?

 

ITSM:

  • Create New Incident
  • Check Incident Status
  • Outlook Troubleshooting
  • Catalog Item Request
  • Password Reset

HR:

  • Create New Case
  • Check Case Status
  • PTO Related Inquiries
  • Travel Expenses
  • Update Employee Information

CSM

  • Manage Order Operation
  • Order Case Self Service

 

Q: What ServiceNow patch level is required to leverage the Voice Capability?

A: Zurich Patch 4. Available in Yokohama Patch 11 in January

  

Q: How do I get started with creating my first AI Voice Agent?

A: There are three core steps in the configuration process. They are completed within the Assistant Designer and AI Agent Studio. 

 

JustinWilburn1_0-1765499910564.jpeg

 

Creating an AI Voice Assistant

Creating an AI Voice Agent

 

Q: What license is needed to utilize AI Voice Agents?

A: Generally speaking, you will need a Now Assist license (Pro+ or Enterprise+) to use AI Voice Agents. Please speak with your account team for full details on the entitlements.

 

Q: What drives the questions that the Voice Agent asks during the conversation?

A: The conversation flow is driven by our Voice Orchestrator. End user utterances are routed to the LLM, which generates the next meaningful response and determines where the conversation should go.

 

This behavior is influenced by the user’s input, the AI Voice Agents configured for that service, and the tools available to those agents—such as scripts, flow actions, subflows, and RAG-based search.

 

Q: How many assists are consumed per call?

A: 50 assists per call is the current consumption rate. However, we strongly advise checking the Now Assist Overview publications regularly for the most current rates.

Please Note: Calls with a duration less than 30 seconds will not consume any assists.

 

Q: What languages are supported in AI Voice Agents?

A: In addition to English, the Dec. 2025 store release adds support for German and Spanish. Additional languages will be supported in future releases

 

Q: Can I access call transcripts?

A: Yes, the full transcript from each call (caller and voice agent) gets logged within an Interaction record and Conversation table as soon as the call ends

 

Q: What model provider(s) are we using?

A: We support Azure OpenAI, Gemini, and Anthropic Claude

 

Q: Does the AI Voice Agent support channels outside of IVR? For example, omni-channel experiences such as portal, mobile app, or even collaboration platforms like MS Teams?

A: Callers will be able to engage AI Voice agents from within the SN mobile app, starting in Q1.  We will support other channels later in 2026.

 

 

CONFIGURATION

 

Q: What fallback options are available?

A: When a call exceeds an established time limit (10 minutes), or the AIVA runs out of assistance/answers to offer, fallback is initiated. Current fallback options include:

  1. Connect to a live agent.
  2. Generate a ticket with Record Producer

Q: Will Voice AI Agents leverage the existing (chat based) AI Agents?

A: Not yet. In the initial release, AI Voice Agents operate separately from the AI Agents used by Virtual Agent. They can, however, use many of the same tools—such as record operations, subflows, scripts, flow actions, RAG search, file upload, and MCP server—but the voice runtime is optimized specifically for real-time interactions.

Longer term, our direction is to converge these runtimes so a single AI Agent framework can power voice and chat.

 

Q: Will KB articles be summarized and read back within the voice channel?

A: Yes, KB Summarization and KB Search can be achieved through the OOB Search Retrieval Tool.

Note: To optimize response time, filter to the specific knowledge bases that address common use cases to the Voice Channel (rather than an open-ended search of all KB's).

 

Q: What type of authentication is supported? Can platform data be used in authentication?

A: Knowledge Factors is a supported authentication method, and it uses information stored on platform (e.g. Data on the User table) to aid in authentication. The following is the full list of supported authentication methods in AI Voice Agents:

 

  • PIN
  • SMS Code
  • Knowledge Factors
  • OTP (Auth App/One Time Passcode)
  • Okta Push Notification

 

Q: While configuring the Voice Assistant, the 'Voice Identification' selections (within Authentication tab) are empty?

A: This means that no Knowledge Based Authentication factors have been configured on your instance. A common example is a question record for Phone Number, and an answer record referencing a field (with phone number field type) on the sys_user table.

 

🔍Refer to Step 3 of the Instance Setup Guide, attached down below, for explicit instructions on how to set up Knowledge Based Questions & Knowledge Based Answers. 

 

Q: Will Voice interactions be summarized for live agents (similar to chat summarization from the Virtual Agent)?

A: Yes, we summarize the transcript upon live agent transfer and that is captured in the Interaction Record.

 

Q: Regarding Analytics, what experience will be available?

A: An AI Voice Agent Analytics dashboard aggregates key metrics related to voice assistant usage and performance - including total voice conversations, deflected conversations, conversation outcomes, satisfaction scores, and performance of AI agents invoked by voice assistants.

 

You can access the dashboard from the Analytics tab within Assistant Designer

JustinWilburn1_1-1765500641682.png

 

Q: Is there regulated market (GCC, etc.) support for the Voice capability?

A: Not yet, but there will be soon. GCC support is planned for Q1. With other regulated markets supported targeted for Q2 ‘26.

 

Q: Can AI Voice Agents be activated for a given option in the IVR call tree? Say 'press 1 for password reset,' and then SN would pick up that call.

A: Yes, depending on the telephony setup, there are two common options:

 

  1. Global line with multiple departments, which may include IT, Customer Service, Facilities, Finance, etc. For this use case, keep the CCaaS provider's IVR tree, and have the defined option route to ServiceNow Voice.
  2. A dedicated Global IT or Global HR line - Remove the IVR tree and go directly to Now Assist in Voice

 

CONTACT CENTER (CCaaS)

 

Q: What Contact Center as a Service (CCaaS) providers are supported by Now Assist for Voice?

A: Genesys Cloud and Twilio. Other providers (including 3CLogic, Amazon Connect, and Five 9) are planned for future releases.

 

Q: How are CCaaS Providers integrated with ServiceNow?

A: Configuration begins within the Assistant Designer on the Now Platform and concludes within the Contact Center Application. See the ‘Integrating voice service with CCaaS providers’ Product Documentation for details

 

 

Still have questions? Check out the Full AI Voice Agents FAQ Below

 

Comments
MagnusSchlamovi
ServiceNow Employee

Hi @Justin_Wilburn . Thank you for this elaborate overview. Where in the Now Assist Overview is the consumption rate for AI Voice agents stipulated? I assume they are classified as a medium-sized agentic workflow, but Voice Agents are not mentioned.

Travis Rogers
ServiceNow Employee

@MagnusSchlamovi Good callout - That latest version (September 2025) of the NA Overview doesn't yet mention the new Voice Agent capability. That document is updated slightly out of sync with product release timelines. When the newest iteration is posted, it should clearly call out Voice Agents as a separate AI skill type. 

Mdsaqib826
Tera Expert

Hi @Justin_Wilburn @Travis Rogers @MagnusSchlamovi 

 

This is a great overview of AI Voice Agents and real-time transcription.

I have one clarification question:

 

Does ServiceNow currently provide any native capability to parse or transcribe a standalone WAV audio file (for example, a voicemail or attached audio file on a case) into text, outside of live voice interactions and calls?

 

If not, is there any roadmap or recommended OOB approach using Now Assist / AI capabilities for converting attached audio files into text within ServiceNow?

karthiknara
Tera Contributor

karthiknara_0-1767637212652.png

 

The primary method drop down is empty for the caller identification, it does not have any value and not able to proceed even with the 'Is your primary method a phone number?' option checked. Can you suggest how to fix this?

Travis Rogers
ServiceNow Employee

Hey @karthiknara - You'll need to first establish your methods for user identification under All > Authentication Factors > Knowledge Based Factor. Follow the steps here to get that configured. 

 

And @Mdsaqib826 that specific capability isn't in the immediate roadmap that I'm aware of, but you could handle this with a flow-based API calls to a LLM provider's speech to text transcription model (examples: OpenAI, Gemini). You'll need to be mindful of the file size - OpenAI currently only supports up to 25MB audio file, and Gemini requires upload to their Files API if the payload + audio file is larger than 20MB. 

Rahul Chand
Tera Contributor

Hi @Justin Wilburn1 @Travis Rogers @MagnusSchlamovi ,

While setting up a voice agent assistant. I was able to setup the authentication process successfully. However, while configuring the telephony provider in both the Twilio application and Genesys, the auto-generated URL is not being created.

RahulChand_0-1769514688891.png

 

I followed the ServiceNow documentation shared in the link as well.
https://www.servicenow.com/docs/r/intelligent-experiences/integrate-voice-service-with-twilio.html

RahulChand_1-1769514783584.png

 


Still after refreshing the connection of the AIVoiceAgents record in sys_service given in the documentation, I am unable to obtain the auto-generated URL that needs to be configured in Twilio and Genesys.

Brian Bakker
ServiceNow Employee

@Rahul Chand 

When I clicked on the "Refresh Connections" button on the [sys_service] record for AIVoiceAgent, it populated the URL in the Now Assist Voice Deployment configuration in Assistant Designer. Does the associated [sys_service_endpoint] record have a URL and a sys_id in the Authentication field? DISH should populate this endpoint record with all the details, when you click on the "Refresh Connections" button.

Travis Rogers
ServiceNow Employee

@Rahul Chand what you experienced was a known defect at one point. Not sure if that was patched yet or not. As a workaround, in Twilio you can use this webhook URL (with your instance name replacing the MYINSTANCE placeholder) 

https://aivoiceservice-dart-main-dart.sncapps.service-now.com/twilio/voice/agent_operator?instance_n...

Martin Rudack
Giga Sage

@Rahul Chand  Only using the correct URL on Twilio side while the URL is still "-" in the Assistant Designer will not fix the problem.

This is because once the Voice Orchestrator tries to create a conversation using the AI Voice REST API on your instance, the configuration of the assistant will be checked and fails because it is still “-“.

I had the same issue. I wrote an article about my troubleshooting experience and how I fixed it. You can read it here: https://www.servicenow.com/community/servicenow-ai-platform-blog/troubleshooting-my-first-ai-voice-a...

 

Sivaramakrishn3
Tera Expert

@Justin_Wilburn  Is multi language support limited to one language selection at the moment? Or the agents can switch between any of the supported languages while speaking to a caller?

 

Sivaramakrishn3_0-1770170554305.png

 

Justin_Wilburn
ServiceNow Employee

@Sivaramakrishn3 - You configure the language at the Voice Assistant layer. So yes, one language per voice assistant. This aligns to the feedback we received from customers on their CCaaS setups today. 

Justin_Wilburn
ServiceNow Employee

@karthiknara  - If you didn't yet find an answer to the empty Voice Identification selections, we added a question to the FAQ above (under Config).

 

Even better, we attached the Instance Setup Guide to this article. And within it are specific steps to establish KBA factors, which is the missing piece you were searching for. 

karthiknara
Tera Contributor

Thanks Justin, I configured the identification parameters. But the Genesys URL remains empty, i have verified the sys_service configurations and also tried refreshing. 

JagSyd
Tera Contributor

The link shared in theline takes to ServiceNow main website and not an documentation.
A supported Contact Center Integration with systems such as Genesys Cloud or Twilio. Additional details can be found

karthiknara
Tera Contributor

Provider URL is a known issue and fixed in 2.1, until then the script include can be updated as in the KB Admin users are not able to see the URL retrieved from the sys_service_endpoint records - Known Erro...

Version history
Last update:
‎01-15-2026 11:30 AM
Updated by:
Contributors