@Marcos Gianoni - Thank you for your feedback and question! You’ve brought up a critical challenge . . . the pace of AI tooling is accelerating, often faster than an organization’s data architecture can keep up. Rather than treating that as an either/or decision, I believe the smartest route is “parallel tracks”: deliver early value where risk is low, while simultaneously investing in data foundation and governance.
On one track, pick low-risk, high-value use cases (e.g., internal knowledge-base search, FAQs, non-sensitive automation) that can benefit from AI now. These help build user trust and demonstrate tangible benefit without exposing critical data. On the other track, formalize your data strategy and governance: establish data owners, standardize and validate core data (CMDB, people, location), define classification and access policies, and build processes for ongoing data hygiene.
In parallel, establish governance around AI itself: create a cross-functional body (e.g., an AI Center of Excellence + model governance committee) to set policy, vet use cases, and manage risk. This I would say "early but cautious adoption with disciplined data and AI governance" approach, lets organizations get some value from “shiny new tools” now, while steadily strengthening the foundations that will sustain long-term success.
I believe that approach balances the competing pressures, delivering value without sacrificing data integrity or long-term trust.
Here is a sample of low-risk/high-value use case you can start with (using Moveworks & ServiceNow):
Self-Service IT / Employee Service Requests (Routines & Repetitive Tasks)
Use an AI agent to handle simple, repetitive IT service-management tasks: password resets, software access requests, common incident submission, standard ticket routing, and basic troubleshooting.
These are “low stakes” tasks: the impact of a minor error is limited, and human oversight can be retained if needed. This makes it a lower-risk way to pilot automation while building trust.
Meanwhile, because such automations use existing identity/CMDB/asset data, this use-case helps expose data quality gaps (e.g. missing attributes, outdated records) in a real-world context. That insight is valuable feedback for the data-governance track.
