All-In on AI: How Smart Companies Win Big with Artificial Intelligence

Summary of All-In on AI

1. 1

   The AI Revolution: Transforming Business Through Intelligence

   Start Listening

   Imagine a world where businesses don't just use artificial intelligence-they're completely fueled by it. Where companies don't implement isolated AI projects but integrate AI deeply into their strategy, operations, and culture. This isn't science fiction; it's happening right now at organizations that have gone "all-in-on-AI." Tom Davenport and Nitin Mittal's groundbreaking book explores this transformation, drawing from their extensive experience working with AI-focused companies worldwide. Davenport, whose Harvard Business Review article on "competing on analytics" was designated one of twelve must-read pieces in the magazine's history, and Mittal, who leads analytics and AI at Deloitte US, offer unprecedented insight into how organizations are achieving remarkable results through comprehensive AI adoption. The book has become essential reading for executives across industries, with Bill Gates naming it one of his top recommendations for understanding how AI is reshaping business competition.

2. 2

   The All-In AI Organization: A New Competitive Reality

   Start Listening

   What exactly does it mean to be "AI-fueled"? While tech giants like Google pioneered the approach of putting AI at the center of everything they do, organizations across traditional industries from banking to healthcare are now applying these same objectives to their specific contexts. These AI-fueled organizations comprise less than 1% of large companies today, but they consistently outperform peers with more effective business models, better customer relationships, and more desirable products. The components of being "all-in on AI" are multifaceted but clear. First, these organizations deploy AI broadly across the entire enterprise with multiple use cases spanning all functions and processes. Unlike companies that might have a handful of AI systems, truly AI-fueled organizations often have hundreds of deployed applications using various AI technologies-machine learning, logic-based systems, and semantics-based AI-often combining multiple approaches for the same application. Second, they excel at getting systems into production where they deliver actual economic value. While many companies struggle to move beyond pilots and proofs of concept (a 2021 IBM survey found only 31% of companies had actively deployed AI in business operations), AI-fueled organizations overcome implementation obstacles by planning for deployment from the beginning and ensuring data scientists work closely with business stakeholders throughout the project lifecycle. Third, these organizations fundamentally reimagine work processes with AI as a central element. Rather than simply automating existing processes, they redesign workflows for the "Age of With" where humans collaborate with smart machines. DBS Bank exemplifies this approach, using AI to dramatically transform anti-money laundering efforts and customer service operations. Fourth, AI-fueled organizations ensure widespread organizational fluency in the technology. Companies like Airbus have retrained over 1,000 employees in AI skills, while DBS Bank has trained 18,000 employees in data skills, with 2,000 becoming proficient in advanced data science. This widespread knowledge enables them to implement AI across all operations. Fifth, they make substantial, long-term commitments involving hundreds of millions or billions of dollars. This means using data and analytics for most decisions, changing customer interactions, embedding AI into products and services, and automating many processes. While CEO advocacy is crucial, commitment must extend throughout all management levels. Sixth, they manage data as a strategic asset, with cloud-based, centralized environments that combine internal and external sources. Beyond good data management, AI-transformed businesses develop unique or proprietary data sources to differentiate themselves from competitors, with some companies creating new business models specifically to access more data. Finally, they establish frameworks for ethical and trustworthy AI to avoid potential losses. While tech organizations have led in establishing formal AI governance mechanisms, many available frameworks can help establish principles for responsible implementation. These organizations create value through multiple levers: accelerating decision-making, reducing costs, finding patterns in complex data, transforming stakeholder interactions, fueling innovation with new products and business models, and building trust through risk management. The most advanced companies function as organizational learning machines, continuously improving through experimentation and adapting their AI models as the world changes.

3. 3

   Leadership and Culture: The Human Foundation of AI Success

   Start Listening

   Beyond technology and data, human factors like leadership, culture, attitudes, and skills critically influence an organization's AI capabilities. Despite large technology budgets, many companies struggle to become more data-driven, with surveys showing the percentage of organizations claiming data-driven cultures has actually declined from 37.8% to 31.0% between 2017 and 2021. This decline highlights the challenge of cultural transformation, even with substantial technological investment. Piyush Gupta, DBS Bank's Group CEO since 2009, exemplifies transformative AI leadership. He turned a bank once mockingly called "Damn Bloody Slow" into an award-winning digital banking powerhouse through several strategic initiatives. He emphasized early experimentation by setting an ambitious KPI of 1,000 experiments annually, empowering business units to hire their own data scientists, and establishing innovation labs across Singapore, India, and Indonesia. Unlike most CEOs who delegate digital transformation, Gupta personally led DBS's data transformation, drawing on his experience creating Citi's first data centers in Asia. He focused intensively on talent development, building a team of 1,000 data professionals while implementing a comprehensive data literacy program across all organizational levels. Effective AI leaders share several key traits that distinguish them from traditional executives. Technical familiarity provides a significant advantage, though leaders without this background can learn enough with sufficient effort - as demonstrated by executives like Jeff Bezos who invested time to understand AI's fundamentals. Working across multiple organizational fronts is crucial - actively signaling interest in AI initiatives, establishing data-driven decision cultures, promoting innovation through dedicated funding and resources, and motivating employee skill development through visible support and incentives. Leaders control crucial financial resources, with many initially funding AI experimentation without demanding immediate ROI, recognizing that "ROI too early kills experimentation." This approach has been successfully implemented at companies like Microsoft, where Satya Nadella allocated significant resources to AI research before clear commercial applications emerged. Creating a culture that embraces data-driven decisions and AI's transformative potential is particularly crucial for legacy companies with established ways of working. Many implement comprehensive data literacy programs teaching employees about data types, analytics applications, and data-driven decision making. Effective programs combine conceptual learning with experiential components like simulations and case studies, followed by ongoing reinforcement through mentoring and project-based learning. Companies like Mastercard have created internal "data universities" with multiple learning tracks tailored to different roles and expertise levels. For AI function leaders, "evangelytics"-evangelizing analytics' business value-may be their most crucial role in driving organizational change. Bank of Montreal's approach demonstrates strategic focus, initially targeting data-rich business units like retail banking and wealth management while moving more cautiously with conservative divisions like corporate banking. Successful AI leaders also secure visible executive support through regular board presentations and executive workshops, communicate results frequently through showcase events and internal case studies, and balance short-term wins with long-term transformation narratives that maintain momentum and support. Proactive organizations believe preparing employees for AI is essential for long-term success. Despite automation moving slower than predicted, 82% of AI adopters expected substantial job changes within three years. Companies like Amazon committed $700 million for retraining, focusing on distribution center workers and nontechnical staff through programs like Machine Learning University and technical apprenticeships. DBS Bank implemented "DigiFY" to teach seven digital skills including design thinking and data analytics, and created "translator" roles to bridge business and technical teams. Shell partnered with Udacity to train over 500 employees in AI fundamentals and advanced applications, while Airbus developed a comprehensive AI Academy that trained 1,000+ staff in data science, resulting in over 300 AI projects across the organization.

4. Chapter 4

   Strategic AI: Creating New Value Through Intelligence

   AI strategy requires conversations among senior leaders about how it can improve business, create growth, and generate revenue. Organizations typically pursue one of three strategic archetypes: creating something new (businesses, products, models), transforming operations for efficiency, or influencing customer behavior. The first archetype-creating something new-enables all-in companies to develop entirely new business approaches through new markets, products, services, and business models. Loblaw, Canada's largest grocery chain, demonstrates how AI can fuel expansion into new markets. After acquiring Shoppers Drug Mart, QHR (electronic medical records), and investing in telemedicine provider Maple, Loblaw developed the PC Health app-an AI-powered platform that helps Canadians navigate the healthcare system while integrating with their loyalty program. In product development, Toyota offers a promising alternative to fully autonomous vehicles with its Guardian system, focusing on making human driving safer rather than replacing drivers. Unlike competitors pursuing full autonomy, Toyota's approach aligns with its culture of gradual, reliable improvement. Guardian uses "blended envelope control" where computers can override dangerous driver inputs, similar to fighter jet systems. For services, Morgan Stanley transformed its wealth management offerings by implementing a Netflix-like recommendation engine called Next Best Action (NBA). Using machine learning, the system helps financial advisors identify personalized investment ideas for clients in seconds instead of 45 minutes. Though usage is voluntary, advisors who adopt it work more efficiently and maintain better client engagement. AI enables powerful multisided platform business models that command valuations four times higher than legacy business models. Ping An exemplifies this approach, with its ecosystem generating massive data assets-over thirty thousand diseases and more than a billion medical consultation records-creating what their chief scientist calls "a deep ocean of data." Their ecosystem approach delivers tangible business results: 36% of Ping An's 37 million new customers in 2020 came through its ecosystems, and nearly 62% of its 223 million retail customers used health-care ecosystem services by mid-2021. The second strategic archetype-transforming operations-helps companies make existing strategies more successful. The Kroger Co.'s AI strategy extends beyond its initial implementation, with its 84.51 analytics unit powering Kroger Precision Marketing, which leverages purchase data from 60 million households across 2,800 stores to create targeted campaigns. The company delivers eleven billion personalized recommendations weekly and has partnered with UK vendor Ocado to bring robotics-based fulfillment centers to the US. The third archetype-influencing customer behavior-is inspired by the success of companies like Google, Facebook, and TikTok in changing purchasing, socializing, and information consumption behaviors. Progressive Insurance computes driving scores through telematics data, while health insurers like Manulife and John Hancock use machine learning to monitor and influence health behaviors. These scoring systems, which rely on voluminous data and personalized scoring processes, would be impossible without machine learning.

   Chapter 5

   Technology and Data: Building the AI Foundation

   While human aspects of AI are most critical, companies can't achieve greatness without extensive use of AI technologies and substantial data. AI-fueled organizations pursue clear business objectives through their technology initiatives: supporting diverse AI use cases with broad toolkits, building applications faster with tools like automated machine learning, achieving wide-scale AI deployment, managing data for model training, addressing legacy applications, building high-performance computing infrastructure, and improving IT operations with AI itself. DBS Bank exemplifies the use of diverse technologies across approximately 150 AI projects. For financial crime prevention, they combine rule-based systems with machine learning to reduce false positives by creating risk scores for suspicious cases. Their Transaction Surveillance group developed network link analysis using graph databases and machine learning to detect fraudster networks. DBS employs various machine learning types-neural networks for credit decisions, deep learning for image and speech recognition, and traditional models for predicting ATM cash outages. Organizations embracing AI are seeking ways to develop algorithms more quickly. At Kroger's 84.51 subsidiary, they're developing a "machine learning machine" to build and deploy numerous models with minimal human intervention, supporting their massive data science operation that analyzes billions of customer transactions and delivers personalized offers at scale. Their factory-like approach to machine learning represents an advanced implementation that other companies will likely adopt in the future. Shell exemplifies both the need for AI scale and methods to achieve it rapidly. Using similar development processes and tools across the organization, they've enabled code sharing and rapid deployment. For predictive maintenance, they've trained over 5,000 engineers who already work with equipment to develop and maintain models. In pipeline maintenance, they use drones with cameras and deep learning models to detect issues, reducing inspection time from years to days. Data is the fundamental prerequisite for machine learning success, and acquiring, cleaning, and integrating the right data remains the biggest obstacle for most organizations in scaling AI systems. AI-oriented companies typically maintain cloud-based, centralized data environments with machine-readable formats that combine both internal and external data sources. Rather than focusing primarily on data capture and storage, they emphasize data consumption and the creation of data products that combine data with analytical or AI models. Integrating AI systems with legacy transaction applications presents a significant challenge for many organizations. Anthem Inc. provides an instructive example of managing this transition. They've been consolidating multiple claims processing systems into a single platform with modular services and AI capabilities. Rather than rebuilding everything at once, Anthem uses three-year plans with clear goals and value metrics-acknowledging that legacy companies can't afford to rebuild everything simultaneously, especially given the rapid pace of AI evolution.

   Chapter 6

   Building Sustainable AI Capabilities

   Building sustainable AI capabilities takes experimentation, time, mistakes, and setbacks-it's truly a journey. AI capability maturity depends on factors including use case breadth, technology diversity, leadership engagement, data-driven decision making, resource availability, production deployment extent, business transformation links, and ethical AI policies. Ping An exemplifies an AI-fueled business that transformed from an insurance company into an integrated financial services platform spanning insurance, banking, healthcare, auto services, and smart cities. Its founder and chairman Peter Ma Mingzhe actively engages with AI initiatives, while co-CEO Jessica Tan oversees AI operations. The company employs over 4,500 data scientists and 110,000 technical experts, led by Carnegie Mellon PhD Jing Xiao as chief scientist. Ping An's extensive AI applications include the Good Doctor platform serving 400 million subscribers, intelligent disease prediction systems for cities, and an Auto Owner app that resolves accident claims from smartphone photos in minutes. Scotiabank demonstrates that organizations can rapidly accelerate AI capabilities even after a slow start. Their "blue collar AI" approach prioritizes practical projects with high business value over experimentation. The bank focuses primarily on improving customer experiences and operations rather than pursuing dramatic business model changes. Their AI-driven marketing engine analyzes customer life events and channel preferences to deliver personalized banking advice. This clear operational focus has enabled Scotiabank to catch up to and even surpass competitors who started earlier with AI. While most companies use AI for operational improvements, some insurance firms are pioneering its use to positively influence customer behavior. Progressive pioneered usage-based insurance with its Snapshot program, which has collected data on over fourteen billion driving miles since 2008. Using machine learning models and AutoML, Progressive analyzes driving behaviors to determine personalized insurance pricing. The program influences safer driving through multiple mechanisms: pricing discounts up to 30%, safety letter grades, real-time alerts when unsafe behaviors occur, trip reports, and ML-generated driving tips. While most organizations agree ethical AI is important, few have established the necessary structures and processes. Most companies with AI ethics policies are tech vendors like Google, Facebook, Microsoft, and IBM. These organizations have created ethics officers who focus on internal evangelism and customer education. Some have developed tools like Google's model cards or Facebook's Fairness Flow to document data sources and evaluate algorithmic bias. Unilever's implementation of AI assurance policies demonstrates the complexity of ethical AI governance in a global company with many external suppliers. Their process examines each new AI application for intrinsic risk, with different use cases requiring different levels of scrutiny. As AI enables granular treatment of customers and employees, Unilever recognizes the fine line between personalization and bias, and acknowledges that both regulatory environments and company policies around responsible AI will evolve dramatically in coming years.

   Chapter 7

   Industry Applications: AI in Action

   AI-fueled companies implement specific use cases across different industries to differentiate themselves from competitors and advance their business strategies. Consumer industries-spanning retail, manufacturing, automotive, hospitality, and transportation-share common challenges around understanding customer preferences, logistics, product development, and customer engagement that AI can address. Despite being born non-digital, Walmart has become one of the most AI-capable consumer businesses. Its supply chain leverages hundreds of data scientists for forecasting and demand management, sophisticated route optimization algorithms for its delivery fleet, and AI models to determine alternative products when items are unavailable online. The company is investing $14 billion to modernize distribution centers with AI and robotics, partnering with Symbotic for warehouse automation and with Ford's Argo AI to pilot self-driving delivery vehicles. While many energy and industrial companies have been slower to embrace AI due to B2B relationships and integration challenges with physical machinery, leading organizations are making significant progress. Seagate Technologies, the world's largest disk drive manufacturer, has extensively leveraged factory sensor data to improve manufacturing quality and efficiency. Their primary focus has been automating visual inspection of silicon wafers through deep learning image recognition algorithms that detect and classify wafer defects-improving inspection accuracy from 50% to over 90% while generating multi-million dollar savings. Financial services-including banking, insurance, investment management, and trading-lead all industries in AI adoption due to their information-rich environment, need for rapid accurate decisions, customer advisory requirements, and substantial investment resources. Capital One, the third-largest US credit card issuer, evolved from an analytical company to an AI powerhouse with machine learning applications across consumer banking. Beyond credit decisioning, they use AI to diagnose app failures, identify money laundering and fraud, create virtual card numbers, predict customer intent and call center needs, and power their Eno chatbot. The US government has accelerated AI adoption following Executive Order 13859, with nearly half of federal agencies experimenting with AI tools. Implementations include NASA's RPA projects achieving 86% automation of HR transactions, NOAA's comprehensive AI strategy, Social Security's machine learning for adjudication, VA's AI Institute and Covid-19 chatbots, Justice Department's crime-fighting AI research, Homeland Security's airport screening improvements, and IRS's AI-optimized taxpayer communications. Life sciences and healthcare companies stand on the brink of AI-driven transformation. Cleveland Clinic uses predictive models to identify high-risk patients, detect disease risks before symptoms appear, and identify patients with problematic social determinants of health. The clinic also employs deep learning for medical image analysis, with radiologists experimenting with automated cancer and bone fracture identification, and neurologists using AI to identify epileptic seizure sources. Major pharmaceutical companies like Pfizer, Novartis, AstraZeneca, and Eli Lilly are implementing diverse AI applications from sales and marketing to drug development and clinical trials. Technology, media, and telecommunications industries are among the most digitally advanced and AI-powered sectors. Disney's AI journey began in 1995 when the Parks and Resorts unit recruited airline yield management expert Mark Shafer to apply dynamic pricing to hotel rooms. Today, the company uses AI across all business units, from the Genie app that recommends attractions based on family preferences to StudioLAB's innovations that monitor audience emotions during test screenings and automatically enhance movie frames.

   Chapter 8

   The Path Forward: Becoming AI Fueled

   For traditional organizations that haven't historically leveraged technology, data, and AI extensively, the transformation to becoming AI-fueled may seem daunting. However, the good news is that no company was powered by AI a decade ago, and the path to aggressive AI adoption doesn't require superhuman traits. Companies that have successfully made this transition saw the need for more AI, put people in charge of creating that future, secured necessary data, talent, and investments, and moved rapidly to build new AI capabilities. Deloitte exemplifies the shift from relying exclusively on human professionals to embracing a collaborative mix of humans and machines. With 350,000 employees worldwide, Deloitte isn't abandoning its human workforce but is making extensive use of AI a hallmark of its professional services. The AI strategic initiative has a five-year horizon (2021-2026) and focuses on both enabling internal capabilities with AI and creating new client offerings. Deloitte's audit and assurance practice began working on AI capabilities in 2014, developing the global AI platform Omnia, which automates audit transactions, prioritizes human auditor reviews, and generates business insights for clients. Capital One, founded in 1994 on the principle of information-based strategy, has evolved from an analytics pioneer to an AI-focused organization. The company appointed the world's first chief data officer in 2002 and built its reputation on data-driven decision making. Capital One has consolidated its machine learning teams into the Center for Machine Learning, with data scientists building models across all business functions from credit cards to human resources. The bank's AI transformation was enabled by cloud migration, beginning around 2011 when executives recognized that declining technology costs, increased digital data, and cloud capabilities could transform customer understanding. CCC Intelligent Solutions exemplifies how a midsize company can transform from data-focused to AI-focused by leveraging extensive data assets and business ecosystems. Originally founded in 1980 to provide car valuation information to insurers, CCC has evolved to manage an enormous dataset of automobile insurance information. The company's AI systems help determine which network participants should be involved in resolving claims, identify applicable local rates and regulations, select optimal repair providers, assess vehicle damage, evaluate potential injuries, and calculate precise resolution costs. Unlike legacy companies transitioning to AI, Well was built as an AI-first startup focused on making people healthier through personalized health recommendations. Founded by Gary Loveman (former Harvard professor and Caesars CEO), Well has raised over $60 million to serve employers, community health organizations, and consumers with AI-powered health advice. Loveman contrasts his experience leading Well with his previous roles at large companies, noting the freedom from legacy technology constraints allows them to build with modern modular software and easily create APIs to other systems. The key lessons from these AI transformation journeys provide a roadmap for organizations seeking to become AI-fueled: have clear objectives beyond financial success; build on existing analytics capabilities; reduce technical debt through flexible IT architectures; leverage cloud computing for data integration; redesign workflows to incorporate AI; accumulate strategic data assets; establish governance structures and leadership; develop AI talent through centers of excellence; make substantial investments; build strong partner ecosystems; and implement solutions across entire organizations rather than in silos. AI-applied strategically and comprehensively-will be critical to business success as data continues to increase exponentially, allowing companies to make smart decisions at scale and transform their competitive position in the rapidly evolving digital economy.

   Continue Reading