Fausto Fleites stands out as a distinctive figure in digital leadership. Serving as the Vice President of Data Intelligence at ScottsMiracle-Gro, he emphasizes a hands-on leadership style. “I’m deeply involved in the development process,” he explains. This direct engagement enables Fleites to maintain a close connection with the practical data challenges businesses face daily. Unlike some executives who become detached from operational realities after promotion, Fleites’ approach ensures he remains intimately familiar with how technology can effectively address organizational issues-an essential perspective in today’s automation-driven landscape.
He elaborates, “My grasp of AI’s capabilities is solid. Many companies misunderstand AI because they either listen to exaggerated claims or rely solely on marketing hype. Success depends on recognizing both AI’s strengths and its limitations, and on leveraging AI as a collaborative tool rather than expecting it to perform all tasks autonomously.”
Before joining ScottsMiracle-Gro in February 2023, Fleites held senior digital leadership roles at Sears and Accenture. The 150-year-old company sought to unlock greater value from its data assets, and Fleites aimed to craft a comprehensive transformation strategy to achieve this.
Strategic Approach to Unlocking Value
Initially, Fleites spearheaded the creation of ScottsMiracle-Gro’s first machine learning applications and established a modern cloud-based data lake on Amazon Web Services (AWS). These initiatives demonstrated how data analytics could drive smarter decision-making-ranging from evaluating sales promotion effectiveness to assessing macroeconomic influences and simulating operational plans.
Following these successes, Fleites was promoted to Vice President of Data Intelligence in May 2024. “My focus now is on optimizing governance of our data and AI assets, and steering AI-driven business transformation,” he shares.
He highlights a pivotal moment in technology adoption: the launch of OpenAI’s ChatGPT in late 2022. “Today, AI adoption is widespread because companies view it as integral to future business models. However, the critical question remains-how do you ensure AI investments yield tangible returns?”
At ScottsMiracle-Gro, AI enthusiasm catalyzed executive leadership to explore emerging technologies as part of their digital transformation journey. Fleites’ expertise in AI and machine learning positioned him to lead these efforts at an appropriate scale.
Fleites outlines a three-phase digital transformation framework: the foundational phase involves crafting a data-driven strategy and securing executive support; the implementation phase focuses on digitizing processes, deploying new technologies, and enabling data-informed decisions; and the realization phase is where measurable business benefits emerge. ScottsMiracle-Gro has made substantial strides in implementation and is now beginning to experience the rewards of this transformation.
“Our automation journey is longstanding, with extensive use of robotic process automation (RPA). Currently, we’re modernizing digital assets by integrating AI and personalization, deploying machine learning models that have already transformed operational workflows. We’re in the implementation phase but are already seeing significant benefits,” Fleites notes.
Integrating Machine Learning for Enhanced Decision-Making
One of Fleites’ proudest achievements has been embedding machine learning deeply into the company’s operations. This integration provides senior leaders with timely, actionable insights that inform strategic decisions.
“Our machine learning applications extend beyond forecasting production volumes. We’ve operationalized explainable AI models that clarify the reasoning behind predictions,” he explains.
Weekly, leaders from sales, supply chain, and brand teams analyze these insights to understand recent trends and adjust marketing investments for upcoming periods. This data-driven approach is reshaping operational tactics across the company.
ScottsMiracle-Gro’s AWS data lake serves as the backbone for these machine learning models. The team employs Apache Airflow for orchestrating workflows and retraining models as needed. Model selection varies by use case: XGBoost and LightGBM are preferred for short-term forecasts, while Google’s deep learning models handle complex store- or region-level predictions involving thousands of SKUs.
Fleites emphasizes that while many large enterprises use machine learning primarily to optimize customer-facing operations like supply chain replenishment, ScottsMiracle-Gro’s approach is unique in delivering granular, insight-driven data directly to senior leadership, enhancing strategic agility.
Harnessing Generative AI for Customer Engagement
Fleites also leads initiatives to introduce cutting-edge AI features for customers. One example is the deployment of Google Vertex AI’s retrieval-augmented generation (RAG) application, which enables natural language search across the company’s product catalog.
Additionally, Google Vertex AI powers AI-driven chat agents that enhance customer service quality by analyzing data insights to address common inquiries and seamlessly escalating complex issues to human agents.
“I now focus on governing our data and AI assets effectively and redefining how we conduct business in an AI-driven world.”
– Fausto Fleites, ScottsMiracle-Gro
This chat functionality is supported by Sierra, a personalized AI customer service platform co-founded by former Salesforce co-CEO Bret Taylor. Fleites’ team integrates product catalog data and customer queries into Sierra via APIs hosted on Google Cloud.
Fleites views these AI-powered search and chat tools as foundational steps. “The next phase involves expanding AI-driven customer journeys, enriching content, and moving beyond initial deployments. AI copilots will revolutionize internal processes, freeing employees to focus on higher-value work.”
One notable internal innovation is the “Email Rewrite” tool, launched about a year ago. Previously, customer service reps manually compiled email responses using Salesforce knowledge articles-a slow process that often resulted in inconsistent messaging. The new tool automatically generates coherent, brand-aligned emails in under 30 seconds, boosting both response quality and volume by up to tenfold. This marked ScottsMiracle-Gro’s first back-office automation success, and Fleites is now exploring further opportunities through an internal initiative called “X-ray.”
“We aim to map processes from a technological perspective and prepare for future AI-driven workflows. Looking ahead five to ten years, as AI agents become more advanced, this approach will be critical for back-office operations in large enterprises,” Fleites predicts.
Advancing Proactive Personalization and Future Outlook
Reflecting on his progress, Fleites envisions a data-centric enterprise within the next two years, focused on maximizing benefits across customer service and internal functions-the final phase of his digital transformation roadmap.
“Our upcoming digital platform iteration will place AI, personalization, and dynamic content at its core. Leveraging predictive analytics, we’ll anticipate customer needs and transform digital and e-commerce interactions,” he explains.
For instance, customers might soon snap a photo and receive expert advice powered by generative AI. Machine learning will enable the company to forecast customer requirements proactively. “If weather forecasts predict high humidity and potential lawn diseases next week, we can proactively inform customers about what to expect and how to prepare,” Fleites illustrates.
On the operational side, Fleites plans to implement at least a dozen agentic automation use cases across the company within 24 months. “Achieving this will validate our approach and pave the way for scaling back-office automation further,” he concludes.
