- Post History
- Subscribe to RSS Feed
- Mark as New
- Mark as Read
- Bookmark
- Subscribe
- Printer Friendly Page
- Report Inappropriate Content
an hour ago
Master's students studying IT strategy recently interviewed me about how enterprises build an AI strategy that actually delivers. One pattern came up again and again from my work with clients in Thailand and around the world. Companies buy AI features and tools, expect magic, and never define success criteria.
Enterprise AI is not magic. On the ServiceNow AI platform, as anywhere else, AI succeeds in four stages. Skip one and the next one fails.
1. Start with a problem, not a product
A strategy is a goal you can achieve and count. For example, reduce incidents by 30 percent or cut the cost of a process. ITSM and ITOM are a good starting point because their goal is already measurable: fewer incidents and less disruption. AI then speeds up processes that already work. It does not invent them.
2. Get the data ready, and name the owner
AI is only as reliable as the data it uses. Before turning on any AI capability, check that the data is complete and current. Then confirm that every data set has a named owner who governs it. In ServiceNow terms, that means the CMDB, incident history and knowledge base the AI will draw on.
3. Govern by risk level
Not every decision should be autonomous. Low-risk decisions can run on their own. Higher-risk data and decisions need a person to approve them. A subject matter expert should always verify what the AI produces before anyone relies on it.
This is where ServiceNow AI Control Tower helps: an inventory of your AI systems, a named owner and risk rating for each, and assessments against frameworks like the NIST AI Risk Management Framework. My advice is to start from the framework and the regulations that govern you, then add your own experience.
4. Prove adoption with KPIs set before go-live
Measure three things:
-
Utilization. If you implemented ten use cases, are all ten being used?
-
Productivity. Is the work actually faster?
-
The business outcome. Did incidents go down?
If you can't answer these, you don't yet know whether AI worked.
Free checklist
I turned these four stages into a free AI Readiness Checklist with 10 checks to run before you buy another AI tool:
The full interview is on TechTalk with Bill:
Over to you
How is your organization measuring whether its ServiceNow AI capabilities are working? Please share your KPIs in the comments.
Bill Martin
ServiceNow Certified Technical Architect and 2026 MVP
