Voice is your most expensive IT support channel. Here's how to make AI handle it with ServiceNow

[aastha14]

45-60% of employees still pick up the phone even when digital channels exist and those calls cost 2-5x more than digital interactions. AI Voice Agents let you change that! This FAQ breaks down some of the most common ITSM questions about deploying voice self-service on ServiceNow: from CCaaS setup and authentication options to conversation design pitfalls and go-live testing. Whether you're just evaluating or actively building, this is your starting point.

AI Voice Agents extend the same self-service logic powering your Virtual Agent into inbound phone calls, 24/7, without adding headcount.

Q: What are AI Voice Agents and how do they fit into Zero Touch IT Support?
AI Voice Agents are an additional delivery channel for conversational self-service. Employees who prefer to call rather than type get the same resolution paths as those using the Employee Center chat - the same ITSM topics, catalog ordering, and knowledge Q&A delivered over voice. They extend the Zero Touch IT Support rather than replacing any existing component.

Q: How are AI Voice Agents different from the ServiceNow Voice Framework?

These are two distinct, architecturally separate capabilities. The ServiceNow Voice Framework (CTI/Embedded approach) is for fulfillers - agents receiving calls in the agent workspace via Openframe. AI Voice Agents are requestor-facing: AI handles the call end-to-end without a human fulfiller.

Q: How do AI Voice Agents differ from NAVA (Virtual Agent / text)?

The core resolution logic (LLM topics, Knowledge Base, catalog) is shared. What changes is the interface layer:

• Input: NAVA accepts typed text; Voice Agents accept audio only (Speech-to-Text)
• Output: NAVA renders text, cards, and rich UI; Voice Agents produce spoken responses (Text-to-Speech)
• Authoring: NAVA topics are built in VA Designer; Voice Agents are configured in AI Agent Studio
• CCaaS required: NAVA has no telephony dependency; Voice Agents require Genesys Cloud or Twilio
• Languages: NAVA supports multiple languages; Voice Agents currently support English, German, and Spanish (as of December 2025)
• Testing: NAVA has an in-platform preview; Voice Agents require real test calls
• Fallback: NAVA falls back to live agent chat (AWA); Voice Agents transfer via phone or create a Record Producer ticket

Q: Why should we invest in voice if we already have NAVA?

45-60% of employees default to phone even when digital channels are available. Voice calls cost 2-5x more per interaction than digital channels. AI Voice Agents handle inbound call volume 24/7 across time zones without adding headcount, deflecting your most expensive interaction channel.

Q: What platform version and licensing is required?

You need Zurich Patch 4 or Yokohama Patch 6 (or newer) and an ITSM Pro Plus or Enterprise Plus subscription with Now Assist for ITSM entitlement. Validate your version at All > System Diagnostics > System Info and your subscription at System Applications > Subscriptions.

Q: Which CCaaS providers are supported?

Genesys Cloud and Twilio are supported in the current release. Amazon Connect is on the roadmap. If your organisation uses a different CCaaS provider, voice is not supported without custom development in the current release. Genesys Cloud is required if you want AWA-based live agent routing as a fallback.

Q: Are there any environments where AI Voice Agents are not available?

As of Zurich Patch 4, AI Voice Agents are not available in Self-Hosted, In-Country, or NSC environments. However, this is actively changing: GCC support is planned for Q1 2026, and other regulated markets are targeted for Q2 2026. Self-hosted environments running NowLLM gained AI Agents support as of Yokohama Patch 6, though Voice-specific availability should be confirmed with your account team. Confirm your deployment type and roadmap eligibility before beginning any configuration.

Q: What Voice Agents for ITSM are available out of the box?

OOB Voice Agents are available across three workflows in Zurich Patch 4:

ITSM (plugin: app-itsm-voice-aia):

• Check Ticket Status — lets employees check incident status, add a comment, or escalate urgency
• Create Incident — creates a new incident from a spoken description
• Outlook Troubleshooting — guides employees through common Outlook issues using KB Search Retrieval
• Catalog Item Request — submits a catalog item request via voice
• Password Reset — self-service password reset via identity verification

Q: What authentication methods are available for caller identity verification?

Four authentication factors are supported:

• Soft PIN - employee enrols a PIN via My Profile; enrollment codes are sent by email
• SMS OTP - requires the Notify plugin and an SMS provider
• TOTP via authenticator app
• Okta Verify push notification

Knowledge-based authentication (e.g., verifying by mobile phone number stored in sys_user) is also available and requires no employee pre-enrolment - recommended for low-friction deployments.

Q: What fallback options are available when the Voice Agent cannot resolve a call?

Two OOB fallback options are available, configured per Voice Assistant:

1) Connect to live agent - transfers the call via the CCaaS platform to a human agent phone queue or AWA configuration

2) Generate a ticket with Record Producer - creates a ticket automatically and ends the call. Live agent transfer is the recommended default for IT support contexts where some intents always require human handling.

Q: What should I test before promoting to production?

For each active Voice Agent, test with real calls:

• The primary resolution path (successful completion)
• An ambiguous input scenario (agent should ask a clarifying question)
• An incorrect input scenario (agent should offer an alternative approach)
• The fallback path (agent cannot resolve; call transfers correctly)

Also test authentication independently for each configured factor (Soft PIN, SMS OTP, Knowledge-based) using a real test user account. Do not use impersonation for authentication testing.

Q: Can I use AI Voice Agents with Amazon Connect?

Not in the current release. Amazon Connect support is on the ServiceNow roadmap. Only Genesys Cloud and Twilio integrations are supported today.

Q. Is there a way to make AI voice less sensitive to the background noise?

Background noise adaptive filtering will be out in May (target)

Q. Are reference variables on a catalog item supported for AI voice?

Currently, reference fields or complex variables on catalog items are not supported.