From Development to Production: Ask ServiceNow's AI Experts Anything

[ryanyong]

Hey everyone, I'm Ryan, an engineer on ServiceNow's Forward Deployed Engineering team. We're a team of experts and builders who've gained deep AI deployment expertise at tier 1 technology companies and research labs. We're certified experts on the ServiceNow AI Platform, with direct access to Product and the ServiceNow C-suite.

What we do
We partner directly with customers facing the most complex challenges to co-innovate agentic AI use cases and ship them to production on the ServiceNow AI Platform.

Why this thread
If you're a ServiceNow customer sitting on AI entitlement you haven't fully put to work, this is your chance to ask the Forward Deployed Engineers building these solutions in the field. Bring your hardest questions about moving from unused credits to AI running in production, and getting the most value out of your existing ServiceNow investment.

Ask us anything. We are the subject matter experts, and we will get you an answer.

We'll be answering live from August 3rd to 7th. Post your question anytime between now and 8/7, upvote the ones you want answered first.

[GreysonS]

Use the guided path rather than picking plugins off a list. ITOM has a Now Assist for Setup experience with an ITOM Configuration console for Event Management that takes you from install through alert automation configuration in one place.

The install is two plugins, ITOM AIOps and Now Assist for IT Operations Management. From there the console walks the sequence: assign evt_mgmt_admin, create an integration account with evt_mgmt_integration so your monitoring tools can push events, install your integrations, review alerts in Express List, then configure alert automations for enrichment, grouping, ignore, delay and incident creation. Assign evt_mgmt_operator to your ops groups when you're ready for them to work alerts.

Start with the Learn the basics section in that console before configuring anything. It covers CMDB considerations and prerequisites, which is where most of the trouble originates.

On pitfalls:

CMDB coverage determines everything downstream. Alert analysis and service impact reasoning both depend on CI relationships and service maps. Stale or partial data produces confident, wrong answers, which is worse than no answer because operators will act on them.

Don't expect the generative AI skills to reduce alert volume. They help operators understand alerts faster. Cutting the number of actionable alerts is the grouping, ignore and delay automation work in that same console. Do both, but keep them separate in your head so you can tell which one moved your numbers.

Anything that summarizes past resolutions is only as good as your resolution notes. If your team closes incidents with "resolved" or "restarted service," you'll get very little back. Sample some recent closures before switching that on.

Plan the operator rollout as part of the project rather than after it. The console has a dedicated section for training the operations team for exactly this reason. L1 and L2 have to trust the output or none of it matters.

One thing worth confirming early: Now Assist for ITOM covers alert-shaped work. Incident summarization and resolution notes generation are Now Assist for ITSM skills under a separate entitlement. Since your focus is Incident and Task, check the Now Assist Admin plugin page to see what you actually have before building a plan around it.

And capture MTTD and MTTR before you turn anything on.

[Andrea H]

We are deploying Now Assist for ITSM and have noticed that there are no skills for Problem Management. Are there any plans to include that in upcoming releases? We're also very interested in cross-ticketing investigation. For example, when an incident occurs using a skill that can correlate not just existing incidents but also changes that have occurred in the last 48 hours. Are there any plans for this type of addition to the incident assist? Last question, when we notice that the AI has returned an incorrect answer, we tell it that it is incorrect and provide the correct information. It does not seem to learn this and returns the incorrect answer again in the future. Are we doing something wrong that it's not learning from our responses?

[Elena Mata]

How do you recommend measuring the actual business value delivered by ServiceNow AI capabilities (Now Assist, AI Agents, GenAI features) against AI Assist consumption and licensing costs? What KPIs have proven most effective to demonstrate ROI beyond simple usage metrics?

[NicholasG799675]

I love this question. The KPIs you mentioned, usage metrics along with hours saved and cases deflected, are all good, but on their own they never tell the whole story. What I like to measure is how much more effective a human becomes because of a given ServiceNow AI capability or agent.

Take HR case deflection for non-critical cases. You implement it, you measure the cases reaching a human expert, and you'd expect that number to go down. That's the obvious metric. But I'd also measure the time a human expert spends on a case, customer satisfaction, quality of resolution, and the rate of repeat similar cases.

Here's why. When the agent triages and deflects the cases it can solve on its own, your experts get to work uninterrupted on the most critical ones. That has a harder-to-measure effect. You get better knowledge transfer through higher quality resolutions, better customer satisfaction because critical tickets get answered faster and more completely, and experts who get to flex their actual knowledge and skills instead of solving the same KB article over and over.

In short, while I can't speak to individualized metrics for you, I would recommend looking toward metrics that show how AI is multiplying your existing workforce and enabling the highest quality of work, in addition to the raw consumption numbers.

[Elena Mata]

As ServiceNow continues to expand AI capabilities, how should customers design their AI roadmap to avoid implementing point solutions that may become obsolete as native AI Agents and platform capabilities mature?