Tech Leaders Move AI Pilots Into Enterprise Transformation
At a panel convened by Fortune, senior technology leaders from Mars Pet Nutrition, Orange, Reckitt, and Saint-Gobain described how they are moving AI from pilots into scaled enterprise capability. Per Fortune, Rahul Shah, global chief digital and information officer at Mars Pet Nutrition, said the company defined its "five big bets" before chasing scale, then progressed from pilots to scale and from use cases to capability. Fortune reports Ursula Soritsch-Renier of Saint-Gobain favors surfacing high-impact use cases from the daily pain points employees encounter, while Nigel Richardson of Reckitt noted that running pilots is quick and easy but building something scalable is far harder. Bruno Zerbib of Orange cautioned against rushing pilots under external pressure, saying "there is no playbook." The discussion emphasized end-to-end workflow integration, measurable outcomes, and employee adoption over isolated proofs of concept.
What happened
Fortune convened a panel of senior technology leaders from Mars Pet Nutrition, Orange, Reckitt, and Saint-Gobain to discuss how enterprises move beyond AI hype into applied, scaled use. Per Fortune, Rahul Shah, global chief digital and information officer at Mars Pet Nutrition, said the company defined its "five big bets" rather than chase scale immediately, then progressed from pilots to scale, use cases to capability, and information to decisions. Fortune reports Ursula Soritsch-Renier of Saint-Gobain favors surfacing opportunities from the daily pain points employees face across the business, and Nigel Richardson of Reckitt observed that running pilots is quick and easy while building something scalable is a different challenge. Bruno Zerbib of Orange said he values pilots but warned against moving fast just to appear to progress, saying "there is no playbook."
Editorial analysis - technical context
Industry-pattern observations show organizations stuck in pilot purgatory typically lack a capability roadmap, measurable success metrics, and integration with operational workflows. Moving from isolated proofs of concept to capability-oriented programs usually means investing in production-grade data pipelines, observability, and change management. For data and ML teams, that shift favors repeatable feature pipelines, monitoring, and metrics that tie model outputs to business decisions over one-off notebooks and demos.
Context and significance
For practitioners, centering projects on end-to-end processes and employee workflows tends to increase adoption and clarify ROI, which helps justify investment in productionization and model operations. The panel's recurring themes, human-centered scoping and measurable outcomes, mirror what distinguishes durable enterprise AI programs from stalled experiments.
What to watch
Useful indicators include the emergence of capability roadmaps rather than ad hoc pilots, standardized metrics linking models to decisions, investment in observability and MLOps, and tools that embed AI into everyday workflows instead of standalone demos.
Key Points
- 1Defining a small set of strategic "big bets" helps convert scattered pilots into repeatable, scalable capabilities.
- 2Centering AI on end-to-end workflows and real employee pain points improves adoption and clarifies the path to measurable ROI.
- 3Leaders frame scaling as a process-reinvention and capability-building problem, not merely a matter of running more pilots.
Scoring Rationale
Practical, on-the-record guidance from digital leaders at four large multinationals on the pilot-to-production gap, timely for enterprise AI teams. It reports panel guidance rather than new tooling or research, so it is a notable industry signal but not a major development; adjusted down from 6.9.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems