With some disruptions, fast followers gain a competitive edge by waiting to see what mistakes the first movers make. But that’s not what we’re seeing with AI: Early adopters are already starting to realize performance gains up to 20% of earnings in as little as 18 to 36 months. They’re building capabilities and confidence that are likely to translate to a sustained, competitive advantage, empowering them to redefine operations and develop new business models. The last time we saw a new technology this powerful was when the Internet arrived in the 1990s. And this time, change is happening faster.

At the same time, some investors and analysts remain skeptical about returns on investments in AI. This may be because reaping value from AI requires more than just simply conducting trials or deploying the technology. More so than previous disruptions such as the Internet or cloud, AI requires changes in business processes. Companies that conduct business diagnostics, set targets for business deliverables, redesign processes, then develop and deploy AI tools, are seeing extraordinary value.

These early successes are leading to greater investment: The number of large companies investing over $100 million to implement AI has more than doubled in the past year (see Figure 1). These investments are spurring companies to experiment in hundreds of different use cases, but our research finds that most of the value today can be found in five core areas.

Figure 1

Companies of all sizes are meaningfully increasing spending on generative AI

Software and product development

The top use cases for generative AI in software development include code generation, documentation, refactoring, debugging, testing, and run and maintenance. Some developer organizations are already saving 15% to 40% on code generation and documentation, and 30% to 50% or more on refactoring, select testing, and debugging use cases by utilizing the specific patterns and rich datasets that exist beyond the code base.

In some companies, AI deployment has served as the trigger to evaluate software development productivity, expanding their focus to more traditional improvement areas including product management, data-driven prioritization, process stage gate discipline, Agile, and QA shift-left efforts.

Intuit, a financial technology platform for consumers and small businesses, has been testing and scaling more than 30 different use cases to increase end-to-end development velocity throughout the company’s software development life cycle with generative AI. By integrating generative AI technology and tools into its development platform, Intuit is improving productivity for product teams (software developers, designers, product managers, data engineers and analysts, technical program managers, etc.). For code generation, the company has seen greater-than-average efficiency gains by tuning its coding assistant tool on Intuit-specific code context patterns and repositories. It has also focused on a set of refactoring tasks to expedite its code base modernization efforts, further accelerating development velocity.

Customer support

Generative AI can do more than automate and optimize customer support; it can also reduce the amount of support needed in the first place. Generative AI’s application in customer support includes analytics to anticipate, deflect, and address potential customer issues; chatbots to expand digital self-service offerings and automate interactions; algorithms to connect customers with the most appropriate representative; and knowledge assistant tools that help agents act more efficiently.

Generative AI can reduce adviser response time by up to 35%, support consultants during the resolution process by managing different sources of knowledge, and improve the quality of results by up to 40%.

For example, one technology and manufacturing company developed two cutting-edge generative AI prototype applications for field services. The company launched a maintenance assist copilot to boost productivity of field technicians performing maintenance and repair operations, and it developed new systems to analyze huge amounts of diverse and unstructured building sensor data and coordinate information and decision making for emergency responders.

Sales and marketing

In sales and marketing, generative AI is deployed in generating dynamic, personalized content, personalized email marketing, social media engagement automation, automated account planning, and advanced training and support. By automating and optimizing these customer interactions, generative AI is boosting the productivity of sales reps and other marketing staff, shortening cycle times, reducing churns, and delivering better click-through rates through hyper-personalization.

One technology hardware company, for instance, is transforming content management by simplifying content creation, automating systems and workflows that synthesize, assemble, and publish content, and adopting generative AI tools for some roles. The company aims to reduce time spent on content by 30%. Pilots have already delivered promising results in a variety of uses, meeting and exceeding this goal.

New products and features

Companies are deploying generative AI in product and feature development to create simpler and more user-friendly products and interfaces, and to deliver greater customization and personalization. For example, in healthcare, AI can quickly analyze patient data and offer personalized care plans. In other industries, generative AI enables voice or text chat interfaces for simpler interaction with products.

Carrefour’s site, for example, offers a generative AI shopping assistant that can generate shopping lists and menu suggestions based on customer information and input. This simplifies the customer shopping experience while making it more engaging.

Back office

Back-office operations are particularly well suited for generative AI improvements, given the vast number of routine processes that are comparatively easy to automate. In the finance function, for example, generative AI can improve the efficiency of drafting internal audit reports, preparing documentation for tax audits, and running custom financial analyses.

Deutsche Telekom has developed a chatbot for its procurement department that is trained on the company’s policies and historical procurement strategies. The chatbot can answer team requests about policy compliance and provide recommendations on vendors, contracts, or fair price for a specific request for proposal. Pilot results across the company suggest that the chatbot could save business users up to 2,000 hours per month and procurement users up to 5,000 hours per month.

Anticipating challenges

Deploying AI is a transformative journey that aims for significant productivity growth, but involves addressing challenges that span technological integration, human adaptation in ways of working, and reimagined business processes.

• Preparing business processes. In deploying AI, companies should avoid automating existing complexity into their operations. To do that, they should fix the processes before automating by streamlining, simplifying, and eliminating unnecessary steps. This frees up energy and capacity as they modernize operations.
• Modernizing data and application environments. Sprawling databases, multiple sources of truth, and complex application environments hinder the rapid deployment of reliable and productive AI. Investing in modernization and data governance before scaling AI applications releases an additional wave of productivity.  
• Finding technology and services support. Companies implementing AI in the cloud and on premises need reference designs, large language model (LLM) recommendations, prompt engineering, and application development support. All of these resources are in short supply because so many technology providers are currently introducing foundation model AI into their own products. Graphics processing unit (GPU) infrastructure, in particular, is in high demand.

Leading an AI transformation

A strategic implementation of AI aligns initiatives with the organization’s business goals. Whether the changes are incremental or transformational, several best practices are emerging.

• Prioritize AI as a way to generate value, from the CEO down. Set clear targets for return on investment (ROI) and hold teams accountable through the budgeting process for delivering savings and creating value.
• Conduct a business diagnostic. Don’t automate bad processes. Invest in mapping out value opportunities and redesigning business processes before automating. Set targets and manage change to improve efficiency as the technology is deployed.
• Define a clear roadmap for use cases. Focus on functional areas with high value potential, such as sales and marketing, customer support, software development, and operations.
• Leverage multiple AI delivery models, including self-service knowledge worker tools (such as Microsoft 365 Copilot), prebuilt commercial AI systems from vendors, and custom AI models, when the need for differentiation and sensitivity of data is high.
• Build shared datasets, AI models, and technology components and platforms to ensure economies of scale across solutions. Improve product management, as well as Agile and DevOps processes, to support high-velocity AI development.
• Develop appropriate risk management, responsible AI, and governance roles, and ensure clear communication and talent strategies for the workforce.  

For every enterprise, the AI journey will take a unique form. But across industries and markets, it’s clear that the dramatic rise of AI is not a passing hype cycle. The strategic and innovative use of AI will play a key role in achieving competitive advantage over the next decade and beyond. Late adopters are out of time, and companies that fall too far behind the curve will find it difficult to maintain or regain their position.