ServiceNow Now Assist for ITSM for Hospitality

Now Assist for ITSM in a hospitality group is ServiceNow's generative assistant applied to the IT service desk that supports hotel properties. It drafts text and suggests next steps, and each draft is reviewed by a person before it reaches a guest or changes a record.

Now Assist in Hospitality IT Operations

Generated text in a hotel’s IT support desk is only worth having where a wrong suggestion is cheap to catch: a draft self-service note for a guest who cannot connect to wifi, a suggested diagnostic step for a point-of-sale fault. The sections below cover guest-facing self-service, operational support, guest communication, knowledge for guests and staff, seasonal demand spikes, voice and accessibility, staff adoption, and then the decisions: which support tasks may use generated text and which stay human-led, a rollout sequence that protects guests and staff, failure patterns to watch for after go-live, and when generative assistance is not the right tool for the desk.

Guest-Facing System Issues and Self-Service

Many guest issues with IT systems (forgotten passwords, billing questions, reservation problems) can be resolved through self-service. A guest without internet access might use a mobile hotspot. A guest confused about a charge can check their itemised bill or receive an explanation. Now Assist can draft self-service guidance that helps guests resolve issues independently without contacting support.

Guest-facing knowledge articles should be clear, non-technical, and action-oriented. Tell guests exactly what to do: “Check box A in the guest portal, click Report Problem, then select your issue type.” Avoid technical jargon or references to systems guests do not understand. Use examples that match common scenarios.

Operational IT Support and Staff Self-Service

Behind-the-scenes IT supports point-of-sale systems, phone systems, security systems, HVAC, and thousands of connected devices. When these fail, guest experience and operations suffer. Now Assist helps operational teams troubleshoot faster by suggesting diagnostic steps and solutions based on similar past incidents. This reduces mean time to recovery and improves operational reliability.

Integration with Guest Communication

When IT failures affect guests, communication is critical. A billing system failure means guests cannot pay bills. A reservation system failure means check-in is delayed. Now Assist can draft customer-facing notifications explaining the issue, estimated resolution time, and workarounds while systems are being fixed. This transparency reduces guest frustration.

Knowledge Base for Guest and Staff Training

Build knowledge articles covering common guest questions (how to use the TV, how to adjust temperature, how to order room service) and operational staff questions (how to reset a PIN, how to override a lock, how to troubleshoot the phone system). Now Assist learns from this knowledge and suggests relevant articles when new requests arrive.

Keep knowledge content current: update for new systems, retire articles about obsolete systems, and refresh with guest feedback. Outdated knowledge articles do more harm than good, so maintenance discipline is essential.

Seasonal and Event-Driven Demand Spikes

Hospitality experiences seasonal demand surges: holidays, conferences, sporting events. Support volume spikes during these periods. Now Assist helps smaller support teams handle spike volumes by automating routine troubleshooting and speeding escalation for complex issues. This allows limited support staff to serve more guests without degradation in quality.

Continuous Improvement from Support Interactions

Track which issues guests report most frequently and which take longest to resolve. Use this data to identify knowledge gaps: if guests frequently report confusion about a feature, improve training or system documentation. If certain systems have chronic reliability issues, prioritise infrastructure improvements to those systems. Support data drives operational improvements.

Voice Interaction and Accessibility

Some guests prefer voice interaction over text, and some guests have accessibility needs that voice support addresses. Ensure your virtual agent supports voice queries in addition to text. Train the agent to recognise common voice patterns and accents so it understands guests with diverse backgrounds and speech patterns. Graceful degradation ensures that when voice interaction fails, text alternatives work reliably. Design the system to handle interaction method switching so guests can switch between voice and text conversation methods without re-entering information they already provided.

Staff Adoption and Resistance

Support staff might worry that AI assistance will eliminate their jobs or reduce their autonomy. Explain that Now Assist augments their capabilities by handling routine research, freeing them to focus on customer service and complex problem-solving. Staff who engage with AI suggestions often appreciate the reduction in tedious work and preference to focus on meaningful problem-solving.

Decide which support tasks may use generated text and which stay human-led

Generative assistance in an IT service desk is useful where the input is structured and the output is reviewed before it reaches anyone. A duty engineer who asks for a list of likely causes for a point-of-sale outage, drawn from similar past records, is well served. A message that tells a guest their room key will not work, drafted by a tool and sent without review, is not. The first design decision for a hospitality group is therefore a split between tasks where a draft is a starting point and tasks where a human must write, check and send the words. The features themselves are laid out in the Now Assist for ITSM guide.

In practice the human-led list is longer than many teams expect. Anything that speaks to a guest about a payment dispute, a safety fault, an accessibility request or a data question should be written or approved by a named person. Anything that changes a record that drives billing or access should be confirmed by the engineer who performs the change. Assistance can be most valuable for the middle ground: summarizing a long incident thread, proposing a knowledge article from a resolved ticket, or suggesting the next diagnostic step.

A rollout sequence that protects guests and staff

1. Pick a few categories of low-risk, high-volume requests, such as device resets for staff or guest connectivity questions, and limit the first release to those.
2. Check which knowledge articles the assistant can draw on. Retired or property-specific articles should be excluded, or the suggestions will describe equipment that is no longer installed.
3. Agree a review standard for every draft that can reach a guest. Name the reviewer role, the checks they must make, and the record they leave behind.
4. Run a shadow period in which engineers see suggestions but do not use them, and compare what they would have done with what the assistant proposed.
5. Widen the scope one category at a time, and only after the shadow results show the suggestions help the engineers rather than adding a verification step.

Failure patterns to watch for after go-live

• Suggestions cite a knowledge article that describes a previous system. The fix is in the knowledge library, not in the assistant, and the library owner should be told the same day.
• Drafts for guests carry the wrong tone for the brand, either too formal for a boutique property or too casual for a business-travel property. Keep a short set of approved phrases for apologies and timelines so the drafts start from the right register.
• Engineers who accept suggestions without reading them can show up as repeated reopened tickets. Review a sample of accepted suggestions each week and look at the reopen rate for those categories. The governance side of approving generated output is covered in the AI governance and control service description.
• Staff in the back office worry that the tool is judging their work. Explain what is recorded and who sees it before the rollout, rather than after the first complaint.

When generative assistance is not the right tool for the desk

A small property with a single outsourced desk that handles a modest number of calls a week may get more from a well-maintained knowledge library and clear escalation rules than from a generative layer. A property where the core systems are still being replaced may find that suggestions describe a platform that is about to disappear. In both cases, improving the library and the routing comes first, and assistance can follow once the underlying records are reliable enough to draw from. A conversational front end for guests is a different decision, covered in the virtual agent guide.

Questions service desk managers ask before approving the rollout

Does the assistant learn from our guest conversations?

Whether it does depends on how the platform and the instance are configured, and on the data-use settings your group has agreed. Confirm those settings with the platform administrator and record them in the rollout approval rather than relying on an assumption.

Who is accountable when a suggestion leads to a poor fix?

The engineer who accepted and carried out the suggestion remains accountable, in the same way they would be for a fix found in a knowledge article. The point of the review standard is to make that accountability clear and to show where the suggestion, the article or the engineer’s check failed.

Can the assistant handle voice requests from guests?

Voice input depends on the channel configured for the property. Test it with the accents and languages your guests actually use before promising it, and keep a text route open for anyone who cannot complete the voice interaction. Background on conversational routes is in a note on the virtual agent for ITSM.