It was fantastic to see so many of our customers at our latest ServiceNow Managed Service
User Group.
We discussed the topic on everyone’s mind today, AI. But the question is no longer whether we should
be using AI – it’s how.
Gartner’s view is that the organisations with the strongest data foundations will pull ahead. Not only does AI underperform on poor foundations, it can also produce confident mistakes – an even bigger risk.
Throughout the showcase of our customer case studies, demos, and best practices, the concept of strong data foundations came to the fore as organisations look to unlock productivity gains from their AI investments.
Let’s explore the main takeaways from an afternoon of lively discussion.
Four principles for getting your CMDB AI-ready
A Configuration Management Database (CMDB) acts like a map of your organisation, helping AI understand context and relationships.
But, for a CMDB to function well and support AI systems it must remain accurate. It’s through this that you can unlock AI's superpower of pattern recognition across relationships.
To get your ServiceNow CMDB AI-ready follow these four principles:
- Start narrow: Focus on the services AI will touch first – get those right, not everything all at once.
- Fix relationships: Individual records matter less than connections. Assess which service depends on which, and who owns what.
- Build the habit: Without ownership and governance, quality degrades over time, and your AI outputs degrade with it.
- Measure as you go: AI needs metrics, just like any other business-critical process. You need to know your CMDB is current and trustworthy before handing decisions to it.
The governance dynamics behind a reliable CMDB
Mark Adams, National Gas Environments & SRE Manager Technology Engineering, shared lessons on a practical implementation of CMDB and IT operations management (ITOM) discovery.
Putting governance in place before scaling was one of the biggest successes National Gas have found so far, by anchoring data ownership to a structured service model with tightly controlled approvals. A smaller internal, focused, team also helped get the project moving more quickly – with ownership at the correct service layer for clearer accountability.
Following the CMDB implementation, National Gas came away with full visibility into the cloud infrastructure, a stronger foundation for incident and change-impact analysis, and a governance baseline to scale from.
To help people overcome the need to trawl through records, another CMDB governance best practice is to use ServiceNow’s CMDB Health Check Dashboard. Demoed by UP3’s Georg Holzer, we saw how organisations can use this feature to proactively spot issues and act quickly.
The metrics of completeness, correctness, and compliance all provide strong signals about the overall health of your data. The platform automatically generates tasks when records fail health checks, assigning them to data owners or groups, and runs on a recurring schedule, removing the need for manual intervention.
Georg’s advice? “Make it simple, make it easy, make it auditable, and ensure people own the data”.
Rebuilding data foundations from the top down and bottom up
Faced with a database containing nearly half a million configuration items (CIs), Paul Mills, Senior Project Manager, shared how rail operator Avanti West Coast rebuilt its CMDB.
A “great data purge” removed over 387,000 redundant configuration items to restore CMDB trust and accuracy. Strong governance prioritised data quality over quantity, and laid stable service-mapping foundations.
The team tackled it in a two-pronged approach. Bottom up: cleaning CMDB data and implementing proper lifecycle management. Top-down: mapping business capabilities, business services, service offerings, and the technical underpinning. The CSDM (Common Service Data Model) was the framework tying both directions together.
The question of data ownership was bubbling up throughout the afternoon. Paul described Avanti’s distributed approach, making its entire IT community data stewards and embedding responsibility across a governance process. As for the hardest part? It’s people, he said, and if you can bring the people on the journey, you can achieve your goals.
Getting AI-ready from the inside out
UP3’s Technical Director, Justin Loftas, shared our journey with AI in ServiceNow and the importance of strong data foundations.
Using the UP3 Managed Service Integration Application, the team work entirely from our own UP3 ServiceNow instance: customer cases are managed and resolved, then pushed in real time to the relevant customer environments, removing the need for ‘swivel-chair’ working. Sentiment analysis flags customer satisfaction issues, before the tool generates automated resolution notes, produces knowledge articles from resolved cases, and suggests case note tone.
The system works because of the structured data underpinning it. By mapping customer accounts, entitlements, contracts, SLAs, products, and service instances, UP3's team can ask natural language questions in Now Assist to get direct answers without building a report, helping the team work more efficiently. The benefits we’re seeing underline the importance of focusing on foundational data.
The gap between AI use and what’s delivering results
During the final presentation of the day, Nav Venkatesh, UP3 Technical Account Manager, left the audience with the parting idea that every implementation must solve a real problem with a measurable outcome.
Highlighting the rise of tokens (the basic unit of data that an AI model uses to read and generate information) as an emerging operational cost measure, this echoes cloud computing costs from a decade ago.
Understanding your token spend, and ROI delivered is a key measure, AI spend on value delivered, not just AI spend.
To find business value from AI, he suggests identifying the highest-friction workflows and matching the AI tool type (generative, predictive, agentic) to the actual requirement.
So, for example, if a virtual agent is deflecting 40% of cases, the real question is: were those cases resolved? Or did the same user call the service desk five minutes later and log another case?
ServiceNow AI. Seema Shad, Partner Technology Advisor at ServiceNow shared the latest updates from ServiceNow, fresh from Knowledge ‘26, confirming that once the data foundations are in place, organisations can start to reap the benefits of ServiceNow AI such as managing an autonomous workforce using AI Control Tower, taking advantage of ServiceNow Otto, the AI engine powering the ServiceNow platform, and the ability to deploy pre-built “workforce specialist” agents, will offer new ways to unlock productivity and free up your people to innovate.
And finally...
Congratulations to National Highways for winning our Most Valued Partner award, which recognises a customer who's worked with UP3 to deliver innovation or an unusual edge case. The six-month NH Notify project optimised operational notifications, enhancing response efficiency and improving road safety through advanced monitoring.
