Matt Griffiths
CTO
Stanley Black & Decker
In an era where artificial intelligence (AI) is reshaping industries, Stanley Black & Decker stands out for its strategic approach to leveraging AI for competitive advantage. With a long history of integrating traditional AI models, the company was well-positioned to capitalize on the emergence of generative AI technologies when ChatGPT made waves in the data landscape in 2023.
Matt Griffiths, Stanley Black & Decker's Chief Technology Officer and Co-Chair of the Boston CDAO Community, has been at the forefront of the organization's data and technology transformation for over eight years. Reflecting on their journey, Griffiths states, "The concept of using advanced models for building business capability was not new to us… and we felt [GenAI] was a really important key in unlocking our ability to mine unstructured data."
Here, Griffiths delves into their AI journey, highlighting key use cases and successes. He also shares valuable insights and advice for data leaders aiming to harness AI to drive business value.
Stanley Black & Decker’s AI Journey
When ChatGPT came onto the scene, Stanley Black & Decker chose a path of innovation rather than restriction, unlike many companies at the time. Griffiths highlights their strategic decision to embrace the technology, stating, “Our theory was that [if we didn’t] we were just going to drive all of our users underground on their personal devices and that we would lose them.” This proactive approach led to the rapid development of their internal GenAI tool, ChatSBD.
Focusing initially on "low-hanging fruit use cases," such as IT and customer support, Stanley Black & Decker quickly recognized the broader potential of GenAI. As a consumer goods, retail, and manufacturing company, they manage vast amounts of unstructured data – estimated to be over 80% of their data – and they saw GenAI as the key to unlocking insights from this data, particularly in the realm of product reviews.
We quickly realized that the real value was in the ability to get access to insights from data that we just couldn't access.”
For years, they collected product review data from platforms like Amazon, Home Depot, Lowe's, and Grainger, primarily utilizing star ratings. With GenAI, they could now analyze the text data, summarize it, gauge sentiment, and develop tools around these insights. Griffiths explains, “That allowed our engineers to test the next version of a product virtually or for a quality and warranty team to understand why we were suddenly seeing an increase in a product being returned to our distribution centers.”
With that tool, we were able to really start advocating and evangelizing how this technology was going to unlock so many capabilities.”
The success of this use case fueled significant support and investment in their AI initiatives. The momentum created a snowball effect, accelerating the adoption and integration of AI technologies within the company. Today, ChatSBD is in its third iteration, serving around 1,500 unique users daily and handling between 15,000 and 18,000 prompts each day.
In their current phase, Griffiths is focusing approximately 50% of his time on how AI can transform the way IT works at Stanley Black & Decker. He says, “How does AI change the way we go about gathering requirements for IT projects? How does AI change the way in which the first version of code is done? The way that testing can be automated? Is there a world where 80% of your production issues are now being resolved autonomously using AI agents?... It's completely turning on its head the way that we've always thought about IT.”
Three Strategic Approaches to Overcoming GenAI’s Biggest Obstacles
Griffiths identified three major challenges in their AI journey: AI governance, building trustworthy AI, and educating the organization. Here, he shares the lessons learned and actions taken that have been crucial to their success.
- Creating an AI Governance Committee: Early in their AI journey, Griffiths, alongside Stanley Black & Decker’s Chief Information Security Officer and Chief Data Privacy Officer, co-chaired an AI Governance Committee to define the ethical use of AI and establish policy guidelines. The “AI blueprint” they created empowered Griffiths’ team to develop capabilities with business partners without needing constant approval. If a use case deviated from the established guidelines, it was brought to the AI Governance Committee for feedback. Griffiths reflects, “In hindsight, this was the most important thing that we did.”
- Building Trustworthy AI through the Right Technology Stack: At Stanley Black & Decker, there is zero tolerance for GenAI tools providing incorrect information. Griffiths stresses, “It's really easy to build an AI chatbot or agent that builds an itinerary for your vacation to Costa Rica. It's much harder to build an AI capability that consistently makes sure that when a customer asks which nails go into my framing nailer that you give the right answer. In that scenario, providing the wrong answer has a really serious implication and someone could get really hurt.”
Griffiths explains that they dedicate approximately 80% of their development time on the underlying technologies that provide observability, explainability, transparency, bias detection and hallucinations – compared to 20% on the use case itself. He believes it is essential to utilize a technology partner and platform that offers these capabilities and then operate varying large language models from that platform.
- Educating the Organization through Storytelling: Despite high adoption rates of their AI tools, some business partners and employees have never used their internal tools. Griffiths emphasizes, “GenAI is easier to understand when you see it working rather than trying to explain it on a PowerPoint slide.” They use demos and storytelling to illustrate how each role can think about AI.
As an example, he shared that CISOs are typically very risk-averse. However, by clearly communicating the benefits GenAI can have on cybersecurity, Griffiths is now working with Stanley Black & Decker’s CISO on seven use cases. He says, “When you start to story tell how GenAI is going to impact a given individual's world in a positive way, I think that's the key to unlocking the engagement that you need.”
Advice for Data & Analytics Leaders
Griffiths offers three key pieces of advice for CDAOs aiming to achieve business value from GenAI.
- You’ll never be AI-ready. Just bite the bullet.
Griffiths emphasizes that the AI landscape evolves faster than any company can keep up with, and CDAOs who wait to be AI-ready will never get there. While data availability, quality, and trustworthiness are major concerns, Griffiths discusses how they are not strict dependencies in the same way they were in traditional AI and machine learning applications. He notes, “A lot of CDAOs get hung up on ‘We need to get our data ready before we start introducing GenAI,’ and that's absolutely not the case. In fact, in many cases, GenAI can help you fix your data issues.”
- Find a strong technology partner.
Griffiths highlights that many technology partners are at a mature stage in their development, offering remarkable capabilities. For those in financially constrained environments, some partners may provide upfront development costs to aid organizations in their AI journeys. However, he cautions that if the GenAI application consumes LLM tokens, there will be a cost over time, but it can be a beneficial way to get started.
When people ask me why this is such a big deal, my response is that for 60 years we taught humans how to speak computer language. But now, we have computers that can speak human language.”
Evangelize how AI is going to transform every aspect of the way that we work.
Griffiths stresses the importance of data and analytics leaders showcasing the "art of the possible" with AI. He states, “You don't need to be a software developer to write software or advanced applications. You don't need to be a data scientist to build really amazing models. You don't need to be a Photoshop expert to create amazing images. Anyone who can converse in a language with a machine can now do those things.” He recommends conducting a roadshow across the business, but regardless of the method chosen, leaders must “communicate, communicate, communicate.”
Griffiths concluded with a well-known phrase, “Today's AI is the worst it will ever be.” Reflecting on the rapid advancements in AI software over the past few months, he remarked that “it’s a bit scary” to consider where it will be in six to nine months. “The snowball has started rolling already. We, as a data and technology leadership community, shape that snowball and guide that snowball.”
If you are a data and analytics leader navigating the AI landscape, consider joining your local Gartner CDAO Community to discuss tactics for addressing challenges with your peers. If you are already a member, sign in to our app to find your community’s next gathering.
By CDAOs, For CDAOs™
Join the conversation with peers in your local CDAO community.
