MicroStockHub/iStock/Getty Images Plus
Follow ZDNET:
Add us as a preferred source on Google.
ZDNET’s Key Takeaways
An increasing sense of unease is spreading throughout the global technology community, as many analysts and thought leaders warn that artificial intelligence may be teetering on the edge of an unsustainable boom—a bubble whose rapid expansion could soon give way to an equally swift collapse. Yet, amid the tension, the savviest digital executives are responding not with panic, but with precision. They are approaching AI deployment through carefully calibrated, tactical strategies that emphasize specific business use cases. By aligning experiments with measurable corporate objectives and securing executive sponsorship, these leaders are ensuring that innovation remains tied to strategic outcomes rather than hype.
This caution is underscored by a recent MIT study, which revealed that a mere five percent of current AI initiatives yield tangible value for their organizations. Such statistics have only deepened the growing fear that the immense excitement surrounding generative and agentic AI technologies may represent unsustainable enthusiasm that could soon deflate. Yet not everyone subscribes to that pessimistic view. Fausto Fleites, vice president of data intelligence at the renowned gardening and lawn-care brand Scotts Miracle-Gro, is actively working to counter this narrative. His leadership demonstrates that when implemented with deliberation and clarity, AI can be a transformative, value-producing force.
Also: Some enterprises have achieved remarkable success with AI—what sets them apart?
Fleites has long cultivated a career focused on transforming raw data into practical intelligence through advanced analytics, machine learning, and artificial intelligence. These cumulative experiences now power innovative deployments across a company with more than a century of legacy. After holding senior roles at major organizations such as Sears and Accenture, Fleites joined Scotts in February 2023, drawn by the opportunity to bring digital modernization to a 150-year-old enterprise. His initial priority was to establish robust technological foundations by implementing infrastructure through major cloud providers—AWS and Google. On this foundation, he and his team built deep-learning models that convert enterprise data into real-time insights capable of enhancing strategic decision-making among company executives.
Also: I unleashed Copilot on my Microsoft and Google accounts — here’s what happened
Over the past year, Fleites has spearheaded numerous exploratory projects designed to test generative and agentic AI technologies across the business. These initiatives have not only yielded concrete results but also sparked broad organizational curiosity about artificial intelligence. “Our team has been extremely tactical,” he explained to ZDNET. “By focusing our efforts on well-scoped use cases that produce quick wins, we’ve been able to foster enthusiasm and gain traction with other departments eager to replicate that success.”
Six Ways to Create Value from AI
At Scotts, Fleites structures his AI strategy around two complementary dimensions. On one hand, there are consumer-facing innovations designed to enhance customer engagement—such as improved search capabilities and conversational chatbots. On the other, he advances back-office transformations that reimagine internal processes, from rewriting customer service emails to streamlining repetitive administrative workflows. From his hands-on experience, he distills six best-practice principles that can help other digital leaders convert ambition into sustainable value.
1. Iterate rapidly and treat failure as a learning tool.
He offers the example of building a prototype for product recommendations using tools like ChatGPT or Google’s Gemini. “We developed the prototype extraordinarily quickly,” Fleites noted, “and immediately discovered the limitations for our business context. But we pivoted just as fast.” This cycle of continuous iteration fosters resilience: success comes not from avoiding missteps, but from learning through them promptly.
2. Start small and rely on expertise.
Fleites warns organizations to resist the lure of vast, multi-year initiatives devoid of clear business benefit. In a domain evolving as swiftly as AI, long-term plans can become obsolete before completion. “Anyone claiming two decades of generative AI experience is exaggerating,” he quipped. The message is simple: agility and adaptation outweigh scale in such a fast-moving field.
3. Anchor AI within firm strategy and measurable KPIs.
Too many companies, he observes, implement AI simply because it’s fashionable. The outcome of such trend-chasing is typically failure. Real success requires a disciplined framework—connecting each AI use case to an overarching business plan supported by concrete, quantifiable key performance indicators.
4. Encourage cross-functional collaboration and inclusivity.
At Scotts, Fleites organizes cross-functional teams operating much like startups—small, nimble, and empowered to address targeted, short-term objectives. The intention is to spark not just technological change but cultural evolution: encouraging employees to become more data-driven, experimental, and cooperative across traditional departmental boundaries.
5. Refine the foundational platforms.
He emphasizes the importance of strong data infrastructure. By investing in systems that ensure accuracy, accessibility, and security, organizations can empower AI to deliver higher-quality interactions. Scotts leverages the company’s century-and-a-half of product insights to inform intelligent chat capabilities, but to do so effectively, that knowledge must be restructured into AI-ready formats—optimized for efficiency and scalability.
6. Focus on change management and demonstrate tangible impact.
Change management is the bridge between innovation and integration. Fleites recounts that since initiating these AI discussions more than a year ago, his team has presented numerous use cases company-wide. Early wins in consumer services have served as proof points, showcasing measurable benefits that inspire further adoption across departments.
Enhancing Customer Experiences
Both the company’s search and chatbot functions exemplify its progress in consumer-focused AI applications. Operating through a Retrieval-Augmented Generation (RAG) model hosted in Google Vertex AI, the search tool allows customers to query the company’s product catalog in natural language. Before the new system, users had to manually enter exact product names—like “fertilizer”—to retrieve results. Now, the system interprets varied phrasing and conversational queries, improving relevance and user satisfaction.
Simultaneously, Scotts employs AI to enhance its online chat experience. Although currently limited to five core conversational journeys, the chatbot already provides reliable product recommendations free from hallucinations, and addresses frequent troubleshooting topics, including lawn and seed care. When queries require deeper expertise, the AI seamlessly escalates the conversation to a live support agent.
Built through collaboration with Sierra—a technology company specializing in custom AI service agents—the chatbot integrates product databases and FAQ material through APIs on Google Cloud. According to Fleites, the system uses intent recognition followed by probing questions to guide customers toward appropriate recommendations. For example, a user might say, “I want to feed my lawn,” prompting the AI to refine the inquiry—taking into account factors like climate and local regulations—to deliver precise guidance. These efforts represent the early stages of a broader mission to develop a natural, conversational interface that enables users to communicate needs as they would with a trusted advisor.
Reinventing Internal Operations
The second pillar of Fleites’ strategy involves turning AI inward—to reengineer the company’s internal operations. He argues that while customer-facing tools may attract more attention, the greatest return on investment for enterprises like Scotts often lies in automating repetitive back-office functions. The goal is not to replace human workers but to liberate them from low-value tasks, empowering them to focus on creativity, strategy, and decision-making.
A practical illustration is Scotts’ “Email Rewrite” service, a generative tool that draws on internal Salesforce knowledge articles to craft coherent, brand-aligned responses to customer inquiries. Within 30 seconds, the system transforms loose text into polished communications written in the company’s distinct voice. Employees can choose among tone variations that best reflect Scotts’ brand identity. The payoff has been twofold: not only have response times dramatically improved, but message quality is described as an order of magnitude higher than before.
The team continues to identify additional automation opportunities through an investigative framework called the X-Ray process. This approach systematically examines workflows to uncover repetitive, manual activities that can be streamlined through agentic AI. By conceptualizing these systems as intelligent assistants rather than replacements, Fleites seeks to strengthen employee trust and enthusiasm. “We need to shape perceptions now,” he explained, “so that our people see AI as an ally helping them perform more effectively, not a threat to their roles.”
In essence, the initiative at Scotts Miracle-Gro demonstrates how thoughtful execution can distinguish sustainable progress from speculative overreach. Where some fear the imminent collapse of the AI bubble, Fleites’ results suggest a more nuanced reality: a phase of maturation in which disciplined experimentation, grounded strategy, and organizational adaptability convert technological promise into measurable performance.
Sourse: https://www.zdnet.com/article/no-roi-in-ai-yet-try-these-six-proven-tactics-for-creating-real-business-value/
