All-in On AI

All-in On AI Plot Summary

Introduction

Picture yourself standing at the edge of a technological frontier, where companies are being reborn through artificial intelligence. In boardrooms across the globe, visionary leaders gaze at horizons filled with possibility while others remain frozen in uncertainty. Sarah, a mid-level executive at a traditional manufacturing firm, watches as her competitors implement AI solutions that predict maintenance failures with uncanny accuracy, optimize supply chains in real-time, and personalize customer experiences at scale. She feels both excited and anxious - knowing her company must evolve but uncertain how to begin the transformation journey. This tension between technological promise and implementation challenges defines our current business landscape. The AI revolution isn't merely about adopting new tools - it's about fundamentally reimagining how organizations operate, compete, and create value. Through intimate portraits of transformation leaders across industries, we'll discover how pioneering organizations have navigated the complex terrain of AI adoption. From banking giants reimagining customer experiences to manufacturers predicting equipment failures before they happen, these stories reveal a common thread: successful AI transformation isn't primarily about technology, but about leadership, culture, and strategic vision. By examining their journeys - complete with setbacks and breakthroughs - we'll uncover actionable insights to guide your own organization's path toward becoming AI-fueled.

Chapter 1: The Essence of AI-Fueled Organizations

DBS Bank, once derisively known as "Damn Bloody Slow," underwent a remarkable transformation under CEO Piyush Gupta's leadership. When he arrived in 2009, DBS ranked lowest in customer service among Singapore banks. Today, it's a powerhouse recognized globally for excellence, winning accolades including "World's Best Bank" by Euromoney. At the heart of this transformation lies an aggressive adoption of artificial intelligence. Gupta's earliest AI efforts were actually failures - deliberate ones he calls "signaling tools" for the organization. In 2013, he partnered DBS with Singapore's A*STAR research organization, signing a three-year contract to explore AI applications. Though none of those initial projects succeeded commercially, they provided invaluable learning. Gupta established key performance indicators requiring a thousand experiments annually, many involving AI, and organized bi-annual events where these experiments were revealed to encourage innovative thinking. The bank's data transformation was equally substantial. DBS moved data from traditional warehouses to cheaper, more flexible data lakes, created new metadata structures, cleaned up eighty million incomplete records, and developed protocols governing data access. Gupta, drawing from his operations and technology background, personally led this data overhaul - unusual for a CEO of a large enterprise. He continues wrestling with data storage questions, adopting a hybrid cloud approach that balances on-premise security with cloud scalability. DBS also created new governance structures, including a Responsible Data Use Committee that evaluates whether customer data collection is appropriate using their "PURE" framework - data should be purposeful, unsurprising, respectful, and explainable. For talent development, DBS now employs around a thousand data and analytics professionals distributed throughout the organization. Gupta even promoted participation in Amazon's DeepRacer League simulation - an autonomous race car game teaching machine learning - with himself competing and proudly finishing "in the top 100" among employees. The story of DBS reveals a profound truth about AI transformation: technology adoption isn't primarily a technical challenge but a human one. The most successful AI implementations begin with leadership vision, create cultural spaces for experimentation and learning, invest in data infrastructure before AI applications, and systematically develop both specialized and broad AI capabilities across the organization. This pattern of leadership, culture, and capability-building before technology implementation will emerge repeatedly as we examine other transformation leaders.

Chapter 2: Leadership Champions: Visionaries Behind AI Transformations

When Morgan Stanley, the world's largest wealth management firm, envisioned a revolutionary approach to client service, it wasn't primarily the data scientists who drove the change. It was Jim Rosenthal, then chief operations officer, who had the initial vision over a decade ago for a Netflix-like recommendation engine to transform how financial advisors served clients. Working alongside Andy Saperstein, head of Wealth Management (now co-president), he championed what would become the Next Best Action (NBA) system. The system they envisioned would do something seemingly impossible: allow financial advisors to identify personalized investment ideas for clients in seconds – a task that previously required forty-five minutes of research. This was no small challenge with advisors typically serving around two hundred clients each. The NBA system now recommends approximately twenty possible ideas daily for each client, covering everything from bond rating changes and portfolio risk assessments to tax planning considerations. The financial advisor maintains control, deciding which recommendations to send to which clients. What's particularly fascinating is how the system evolved beyond its initial investment focus. Morgan Stanley's management team realized that frequent client engagement was the true key to financial advisor success. They expanded the NBA to include a communications platform, which proved invaluable during the COVID-19 pandemic when advisors sent over eleven million messages to clients in just the first two months of lockdown. Though some critics argued AI couldn't handle sophisticated portfolios with alternative investments like art or private equity, Morgan Stanley proved them wrong by developing approaches that work across all client segments. Jeff McMillan, Morgan Stanley's chief analytics officer, emphasizes this wasn't merely implementing a system but creating "a way of doing business" that competitors would struggle to replicate. He credits the cross-functional approach and executives who maintained their vision over many years. As he puts it, "In the end, financial advising is a human-based game. If all the system does is remind clients that the advisor is there and looking out for them, that is often enough." The Morgan Stanley story exemplifies how transformative AI initiatives require champions at the highest levels of an organization who can articulate a compelling vision, secure resources, and sustain momentum through inevitable challenges. These leadership champions don't need to be technical experts, but they must understand AI's strategic potential and create the organizational conditions for innovation to flourish. Their ability to connect AI capabilities to fundamental business needs – in this case, strengthening client relationships – separates successful transformations from failed experiments.

Chapter 3: Strategic Archetypes for AI Adoption

Ping An, now a global financial powerhouse, began as a modest insurance company in 1988. Today, it's a $200 billion revenue giant ranked sixteenth on the Fortune Global 500 list. How did this transformation happen? Ping An embraced an AI-driven ecosystem strategy that completely reinvented its business model. Rather than simply offering insurance products, Ping An created interconnected ecosystems spanning financial services, healthcare, auto services, and smart cities. Its healthcare ecosystem alone serves 400 million users, providing 1.2 billion consultations through an in-house medical team of 2,000 professionals and 46,500 external doctors. It partners with 189,000 pharmacies, 4,000 hospitals, and 83,000 medical institutions. These ecosystems generate massive data - what Jing Xiao, Ping An's chief scientist, calls "a deep ocean of data" - including information on thirty thousand diseases and over a billion medical consultation records. This ecosystem approach creates powerful network effects. In 2020, 36 percent of Ping An's 37 million new customers came through its ecosystems. By June 2021, nearly 62 percent of its 223 million retail customers used services from the healthcare ecosystem. These customers typically maintain more accounts and assets than others. The company continues pursuing further connections between lifestyle financial services and healthcare ecosystems, creating an "ecosystem of ecosystems." Other AI-focused companies are following similar paths. Airbus launched Skywise in 2017, an open data platform that connects over 140 airlines and 9,500 aircraft. Commercial aircraft produce more than thirty gigabytes of data daily, measuring over forty thousand operational parameters. Airbus developed applications like Skywise Health Monitoring, Predictive Maintenance, and Reliability to improve fleet performance and eliminate unscheduled maintenance. Similarly, Shell established the Open AI Energy Initiative with partners C3.AI, Microsoft, and Baker Hughes to make the energy industry more efficient through AI, focusing initially on maintenance reliability. These examples reveal a fundamental truth about strategic AI adoption: the most transformative implementations don't merely optimize existing processes but enable entirely new business models. The most common strategic archetypes include: creating something new (new businesses, products, or ecosystems), transforming operations (dramatically improving efficiency), and influencing customer behavior. Organizations must carefully consider which archetype aligns with their capabilities and market position. The ecosystem approach, while powerful, requires significant investment in data integration, partner relationships, and AI applications that create value across organizational boundaries. Yet for those who succeed, the rewards include unprecedented growth, customer loyalty, and competitive advantage through continuous learning from diverse data sources.

Chapter 4: Building the Technology and Data Foundation

At the heart of Kroger's transformation into an AI-powered retail giant lies 84.51° - its data science subsidiary handling an extraordinary scale of operations. Consider these numbers: analyzing 2 billion annual customer shopping baskets, leveraging 35+ petabytes of first-party shopper data, and delivering 1.9 billion personalized offers in 2021 alone. This data powers Kroger's "Restock Kroger" strategy, which CEO Rodney McMullen described as using "customer science to make space-planning decisions to disrupt shelf, optimize assortment and improve in-stocks." The scale of machine learning at 84.51° is staggering. Their sales forecasting system, for example, creates forecasts for each item in each of Kroger's 2,500+ stores for each of the subsequent fourteen days - regenerating these predictions nightly based on the most recent data. To handle this immense computational challenge, 84.51° developed what Scott Crawford, a data science manager, calls a "machine learning machine" - a factory-like approach to model development and deployment. This transformation began with an initiative called embedded machine learning (EML), which evolved into a formal mission to enable, empower, and engage the organization to better use machine learning. The "enable" component provided infrastructure - servers, software, and data connectivity. "Empower" involved evaluating over fifty tools and selecting R, Python, and Julia as preferred languages, along with DataRobot as their primary AutoML software. "Engage" meant motivating internal clients by demonstrating benefits through proofs of concept and facilitating code sharing. Perhaps most innovative was their development of "8PML" (84.51° Process for Machine Learning), a standardized methodology with three major components: solution engineering, model development, and model deployment. Unlike many organizations that focus primarily on model development, 84.51° recognized that models not deployed provide no economic value. Solution engineering involves framing the analysis and clarifying business objectives against available resources. Model development utilizes AutoML to speed the process considerably, increasing data scientist productivity. Model deployment addresses the often-underestimated challenge of integrating models into production systems. The results have transformed both 84.51° and Kroger. Professional data scientists, initially skeptical about automated machine learning tools potentially devaluing their expertise, discovered these tools actually freed them to focus on higher-value aspects of data science. The company also expanded participation in machine learning by hiring "insights specialists" - people with less technical training but strong business acumen and communication skills - who could use AutoML to develop models with guidance from experienced data scientists. This story illuminates a critical principle of AI transformation: building a robust technology foundation requires not just tools and infrastructure, but standardized processes that democratize AI development while maintaining quality. The most successful organizations create environments where both specialized data scientists and broader business teams can contribute to AI solutions, focusing their scarce technical talent on the most complex challenges while enabling wider participation through automated tools and clear methodologies.

Chapter 5: Developing Capabilities for Competitive Advantage

When Scotiabank, one of Canada's "big five" banks, decided to accelerate its AI capabilities, it took a distinctly practical approach. Rather than pursuing flashy research projects, Phil Thomas, appointed as executive vice president of Customer Insights, Data, and Analytics (CID&A) in 2019, focused on what he called "blue collar AI" - projects with high likelihood of delivering business value in relatively short timeframes. This pragmatic philosophy pervades Scotiabank's approach. The bank integrated its analytics and data functions under unified leadership, with the chief analytics officer and chief data officer both reporting to Thomas. This structure eliminated the friction that often exists between data and analytics teams, allowing them to move rapidly. While the analytics function is centralized, most data scientists are aligned directly with business lines, ensuring that AI initiatives address real business needs. Grace Lee, the chief analytics officer, emphasized this integration: "Digitization has made the entire bank visible in data, and analytics and AI people are not just enablement—we are a part of the new front lines." The results speak volumes: 80 percent of Scotiabank's analytics and AI models are already deployed in production, with just 20 percent pending. This deployment rate far exceeds industry averages, where many organizations struggle to move models from development to implementation. When COVID-19 hit, Scotiabank rapidly developed an application using machine learning to identify consumers likely facing cash flow issues, enabling proactive outreach from branch relationship managers offering personalized advice and support. They also created an AI-driven marketing engine analyzing customer life events and channel preferences to deliver banking advice through customers' preferred communication methods. The data management function, led by Chief Data Officer Peter Serenita, underwent its own transformation. Previously focused on regulatory compliance and risk management (what they called a "protect-the-bank" approach), they pivoted to faster data delivery through what they termed "reusable authoritative data sets" (RADs). These standardized data assets for customer information, transactions, and balances enhanced speed, consistency, and value generation. Unlike many data projects that struggle to demonstrate return on investment, Serenita reports that Scotiabank's RAD approach consistently delivers measurable ROI. Scotiabank's experience offers a powerful lesson: organizations that start later in their AI journey can catch up and even surpass early adopters by maintaining unwavering focus on business value. Their "blue collar AI" philosophy demonstrates that practical applications delivering clear benefits create momentum for broader transformation. By integrating analytics with data management, aligning technical teams with business units, and emphasizing deployment over experimentation, Scotiabank built capabilities that directly enhance competitive advantage. This practical approach creates a virtuous cycle where each successful implementation builds confidence and appetite for more ambitious AI initiatives.

Chapter 6: Industry-Specific AI Applications

Cleveland Clinic, renowned for innovative, high-quality healthcare, has been methodically integrating AI across its operations, with Chris Donovan, Executive Director of Enterprise Information Management and Analytics, shepherding these efforts. Rather than pursuing a single grand AI strategy, the clinic has fostered a community of practice spanning analytics, IT, and ethics departments to facilitate both bottom-up innovation and thoughtful governance. The applications are diverse and impactful. In surgical care, Cleveland Clinic implemented a machine learning-based preoperative risk score for anesthesia, replacing a traditional rule-based approach with a more automated, precise system. In population health, they built predictive models to prioritize care management resources, ensuring patients most in need receive timely intervention. A diabetic patient struggling to manage their condition receives a high-risk score and proactive check-in calls. Another model identifies patients at risk for developing diseases before symptoms appear, enabling preventive interventions. A third model focuses on social determinants of health, identifying patients who might need social services like transportation assistance as much as medical care. Medical imaging represents another frontier. Radiologists in the clinic's Imaging Institute are using deep learning to help identify cancers and bone fractures, while neurologists apply similar technology to pinpoint epileptic seizure sources. The clinic recently partnered with Path AI to digitize and analyze its pathology slide collection, powering AI-driven research and diagnostics across multiple disease areas. The journey hasn't been without challenges. Donovan notes that healthcare data often presents unique difficulties: "Other industries have much more data, and it's more likely to be clean and well structured. Like other hospitals, our data has quality issues, is captured poorly, is entered in different ways, and involves different definitions across the institution." Even a basic measurement like blood pressure can be taken while the patient is standing, sitting, or lying down - each producing different readings recorded in varying formats. This complexity means data preparation must be part of each AI project. In financial services, Capital One exemplifies AI transformation in banking. Though always analytically driven since its 1994 founding, the company has systematically expanded from credit decisioning to applying machine learning across virtually all aspects of its business. Its portfolio includes AI systems for diagnosing mobile app failures, identifying suspicious transactions, creating virtual card numbers for specific merchants, predicting customer intent during online sessions, and even anticipating why a customer might call their contact center before they dial. What these industry examples reveal is that successful AI transformation requires deep domain expertise combined with technological capability. The most valuable applications address industry-specific challenges - whether optimizing surgical outcomes in healthcare or detecting financial fraud in banking. Organizations must balance immediate operational improvements with longer-term strategic initiatives, cultivating both the technical skills to build AI systems and the business knowledge to identify high-value use cases. As AI continues maturing, the competitive advantage will increasingly come not from having the technology, but from applying it to the most consequential problems in ways that reflect the unique contours of each industry.

Chapter 7: Pathways to Becoming AI-Powered

Deloitte, the world's largest professional services organization, is undertaking a remarkable journey from being exclusively people-powered to becoming both people- and AI-powered. This transformation, led by Jason Girzadas, Managing Principal of Businesses, Global, and Strategic Services, represents a fundamental shift for a company founded in 1845 whose primary asset has always been human expertise. The initiative spans a five-year horizon (2021-2026) with dual focus: enhancing internal capabilities with AI and creating new client offerings. "Our AI initiative is rooted in a belief that this can transform our cost structure as well as our capability set," Girzadas explains. "It's more of a transformational agenda than an objective to develop 'table stakes' capabilities that everyone in the industry will have." The approach includes substantial capital for acquiring AI startups and developing new service offerings, particularly in government program integrity and smart factory management. What makes this transformation particularly interesting is how it's being implemented across Deloitte's diverse businesses. In audit and assurance, the team developed Omnia, a global AI platform that automates audit transactions, prioritizes human auditor reviews, and generates client insights about business risks. Jon Raphael, who leads this innovation, follows a five-step process for each AI use case: simplify and standardize the process, digitize and structure it, automate it, apply advanced analytics, and finally implement AI technologies that learn and improve over time. In tax services, the focus is on extracting and categorizing data from client systems that weren't designed for tax compliance. Beth Mueller, leader of Tax Analytics Insights, notes: "Opportunities for using AI in the tax space abound. We focus on applying highly technical tax law to specific facts." Their Intela platform uses AI to classify trial balance accounts, automate calculations, prepare returns, and perform quality checks beyond human review processes. The consulting practice, meanwhile, is pursuing dual strategies: building capabilities and launching new business ventures. They created the AI Academy to train practitioners on business applications of AI and launched ReadyAI, offering clients teams with preconfigured, complementary skill sets to develop use cases with standard AI processes and tools. Another initiative focuses on automating processes involving multiple transactional systems using AIOps. Deloitte's journey illustrates a universal truth about AI transformation: there is no single path to becoming AI-powered. Organizations must adapt their approach to their specific context, capabilities, and challenges. For some, like Deloitte, the journey involves reimagining core service delivery while maintaining human expertise at the center. For others, it might mean gradually enhancing operational capabilities or building entirely new business models. The common elements across successful transformations include clear vision from leadership, substantial investment in both technology and talent, standardized methodologies that balance innovation with governance, and a willingness to fundamentally rethink how work gets done. The organizations that master this balance between technological capability and human expertise will define the next generation of industry leadership.

Summary

Throughout these stories of transformation, we've witnessed how visionary leaders are redefining their organizations through artificial intelligence. From DBS Bank's journey from "Damn Bloody Slow" to global banking leader, to Ping An's creation of vast healthcare and financial ecosystems serving hundreds of millions, these pioneers reveal that AI transformation is fundamentally a human story. The technology enables the change, but leadership, culture, and strategic vision determine whether that change creates lasting value. The lessons for those embarking on their own AI journey are both practical and profound. First, begin with clear business objectives rather than technology for its own sake. Whether improving operational efficiency like Kroger, enhancing client relationships like Morgan Stanley, or reimagining business models like Ping An, successful AI initiatives solve meaningful problems. Second, invest in your data foundation and standardized processes before expecting transformative AI applications. As Cleveland Clinic discovered, even brilliant algorithms falter without quality data. Finally, recognize that AI transformation requires both specialized expertise and broad participation - Deloitte's journey demonstrates how organizations must simultaneously develop deep technical capabilities while democratizing AI development across the enterprise. By embracing these principles, organizations of any size, in any industry, can harness artificial intelligence not merely as a tool for incremental improvement, but as a catalyst for reimagining what's possible.

About Author

Loading...

Thomas H. Davenport

Read more

Related Books by Thomas H. Davenport

Loading...

Predictive Analytics

Eric Siegel

The Power to Predict Who Will Click, Buy, Lie, Or Die

★★★½☆3.7