FRESH TAKES ON BIG IDEAS

Many leaders are on a similar AI quest these days, seeking to move beyond experimentation and toward real value creation. To reach that goal, they must first solve a three-part puzzle: What should we build, what should we buy, and where should we partner?

This framework prompts leaders to ask themselves what they need to own to create competitive advantage, and what they can safely rent from the market. These questions commonly arise in my AI work in the real estate sector, but they apply to organizations across industries. The answers can help companies shape their technology architectures, develop their distinctive capabilities, and generate real value from AI.

Most organizations have started their AI journeys by giving employees access to gen AI tools, launching pilots in individual functions, and developing early agentic systems. But few have taken the right steps to successfully rewire their operations, which starts with determining which domains and workflows can be reimagined to measurably improve outcomes with AI.

Too often, leaders initially focus on technology instead of strategy. Given the speed of AI’s development, many leaders worry about investing heavily in building a custom tech solution that could become an off-the-shelf commodity in just a few months. As a result, they remain in wait-and-see mode, creating a risk of falling behind more aggressive competitors.

But most organizations do not need to own the underlying technologies that make AI work. They can buy access to large language models, cloud infrastructure, data services, security tools, and agent development platforms from established providers. Instead, leaders should focus on building capabilities that differentiate their businesses. These include proprietary workflows and data, governance models, institutional knowledge, customer relationships, and mechanisms that help organizations learn and improve over time. 

They can then seek partnerships in areas that are essential to transforming their operations, such as redesigning workflows, training employees, and embedding AI into day-to-day activities. Even organizations with strong internal technology teams may find that partners with greater engineering expertise and experience leading similar AI transformations can help accelerate the process and avoid common pitfalls. In some cases, partnering with software providers can be more attractive than buying an off-the-shelf solution because organizations retain greater control over their proprietary data, workflows, and ways of working.

“Most organizations do not need to reinvent the underlying technologies that make AI work. Instead, leaders should focus on building capabilities that differentiate their businesses.”

My recent experience working with a real estate services company shows how these choices can come together. The company sought to automate two highly manual processes in its finance function: matching customer payments to invoices and billing. Employees were spending significant time reviewing financial documents, validating information, and manually entering data into company systems.

Rather than trying to build a new AI-driven system from scratch, the company bought foundational technologies and agent development tools from a leading cloud provider. These already included important capabilities such as an orchestration platform, security controls, governance features, and monitoring.

At the same time, we helped the company build capabilities that reflected its own distinctive way of working. Together, we translated employees’ expertise and business rules into AI-powered workflows. The new agentic system can review financial documents, check them against policies, flag decision-making steps and uncertain cases for people to review, and record the results in the company’s financial systems. The company also selected an external partner for implementation support, process redesign, and change management.

Through this combination of building, buying, and partnering, the company established a new operating model in which people and AI work together. Employees focus on activities that require human oversight, while AI agents handle repetitive, rule-based work. As a result, the company has automated roughly half of its targeted workflow steps. Critically, it has also built a foundation that can directly support broader transformation efforts in leasing, maintenance, procurement, field operations, and other domains where human judgment, improved data, stronger controls, and customer trust create competitive advantage.

This example highlights a disciplined approach that leaders can take as they embark on AI transformations. The best organizations prioritize AI opportunities based on three factors: the value at stake, the feasibility of implementation, and the extent to which a capability can serve as a foundation for future transformations.

This deliberate approach can address one of the biggest surprises I’ve encountered while working on AI projects: the amount of complexity organizations accumulate over time. Global companies, for example, often have different rules, exceptions, and ways of working across countries, business units, and even individual teams. Leaders should start by taking a hard look at how work gets done today and where it can be simplified. AI can even help with that effort. Organizations can use it to analyze documentation, training materials, and process maps to better understand current workflows and identify ways to improve them to create value. 

The companies making the most progress with AI today make deliberate decisions about how to reinvent their operations to solve pressing business problems—not simply automate existing processes. By developing a firm understanding of what they want to build and own, what they are willing to buy, and where partnerships make the most sense, leaders can accelerate their AI transformations and achieve meaningful value.

—Edited by Eric Quiñones, senior editor, New Jersey

ABOUT THIS AUTHOR

Ankit Kapoor is a partner in McKinsey’s New York office.