Situation
The business context
A large restaurant company was undertaking a significant enterprise restructuring following several years of declining sales, earnings, and guest traffic, which placed increasing pressure on operating margins. As part of the broader turnaround strategy, leadership established a mandate to reduce G&A expenses while increasing organizational productivity. At the same time, employees were spending significant time on repetitive tasks and searching across a large Microsoft information estate for the data and knowledge needed to perform their jobs. Rapid advances in generative AI created new opportunities to improve productivity and automate work, but the technology was evolving quickly and where it could create meaningful business value was still uncertain. The organization also had prior experience investing heavily in emerging technologies that did not ultimately produce the expected business value, creating a healthy skepticism toward making large technology bets based on promise alone. Simply introducing AI tools would not ensure meaningful business value or responsible enterprise adoption.
Task
The leadership mandate
As a senior IT leader, I was responsible for developing and executing an enterprise AI strategy that would improve employee productivity, reduce manual effort, and give employees more time to focus on higher-value work. That also meant creating a disciplined way to explore a rapidly evolving technology landscape, build enough practical experience to understand where AI could create meaningful business value, and make informed decisions about where further investment was warranted. The objective was to help the company operate more efficiently without relying solely on traditional cost-cutting measures or making large technology bets before their value was understood.
Strategy
The strategic approach
My strategy was to anchor AI adoption in the company’s turnaround and productivity objectives rather than pursue AI as a standalone technology program. The approach combined broad employee enablement with deliberate exploration of emerging capabilities and targeted AI solutions for specific business problems. Governance, training, and shared learning would support responsible adoption, while hands-on experimentation would be directed toward agreed business use cases, prioritizing highly manual or repetitive processes and areas where capacity or resource constraints were limiting the business. Those use cases would provide practical experience with different technologies and delivery models, while larger investments would follow demonstrated business value rather than technological promise alone.
Action
How the work moved forward
- Created a lightweight model for AI exploration: Established a voluntary, cross-functional “Spark Squad” to gain hands-on experience with emerging technologies while preserving capacity for committed delivery. Exploration focused on agreed business use cases, prioritizing highly manual or repetitive processes and areas where capacity or resource constraints were limiting the business.
- Established guardrails for responsible AI adoption: Partnered with the CISO, Security, and Legal to develop a company AI use policy and ensure pilots and production solutions met security requirements. Responsible-use considerations were also incorporated into solution design and system prompts as AI capabilities moved from experimentation into practical use.
- Launched Microsoft 365 Copilot as an enterprise productivity capability: Partnered directly with Microsoft on rollout, demonstrations, training, and practical guidance, then established a Teams-based AI Community of Practice where employees could share prompts, use cases, practices, and lessons learned.
- Applied generative AI to a Marketing capacity constraint: Worked with Microsoft and the Spark Squad to develop a conversational marketing-copy solution using Azure AI Studio and supporting Azure services. The solution used successful historical content and brand-specific context to generate marketing copy tailored to each brand’s voice.
- Evaluated low-code AI for workflow automation: Used Copilot Studio and Power Automate to explore how reasoning-enabled automation could classify and route guest feedback, including whether nontechnical users could build and maintain AI-enabled workflows that would otherwise require professional development resources.
- Applied AI to unstructured guest feedback: Applied Azure Databricks AI capabilities to guest survey comments that could not practically be analyzed at scale, using sentiment and summarization to add context behind numeric guest-experience scores.
Result
The measurable difference
- Established AI as a practical enterprise productivity capability, extending deliberate experimentation into employee productivity, knowledge access, process automation, and operational insight across multiple business functions.
- Improved employee access to enterprise knowledge through Microsoft 365 Copilot, with employees reporting significant value from being able to quickly find, summarize, and synthesize information across the Microsoft environment.
- Accelerated marketing content creation from weeks to days, reducing manual effort, helping the initial Marketing team meet growing content demand within its existing capacity, and leading to adoption by additional brands.
- Nearly eliminated manual guest-feedback routing, transforming a time-consuming process that previously required dedicated full-time attention into an automated workflow requiring only a few hours of human oversight per week.
- Turned unstructured guest feedback into actionable operational insight, using sentiment analysis and summarization to make large volumes of qualitative feedback practical to analyze and demonstrating enough value to secure additional investment to productionize, improve, and expand the capability.
- Demonstrated that AI-enabled workflows did not always require professional software development, using low-code tools to show how nontechnical users could build reasoning-enabled automation for appropriate use cases.
Leadership insight
Emerging technologies do not require large investments before their value can be understood. Leaders can create space for disciplined exploration around real business problems, learn through practical use cases, and increase investment as evidence of value grows. That evidence will not always translate neatly into hours saved or dollars returned. Adoption, employee feedback, and observable changes in how work gets done can also provide meaningful evidence that a capability is creating value, without pretending the impact can be measured more precisely than it actually can.