A common question I’m hearing from leaders is “how can I best use AI in my business?” They expect concrete, practical answers — not the platitudes and hype that dominate the media.

It’s a fair question: businesses stand to gain significant advantages from judicious use of AI. But first, they must understand where opportunities lie. That requires seeing the business through a different lens.

One way to understand a business is through the value it delivers. For example, a grocery store allows consumers to buy diverse foodstuffs and other household goods conveniently and at reasonable prices. That’s the most obvious lens. But another, just as crucial, is how information moves through the business to support decision-making.

All businesses acquire, process, analyze, communicate, and store data, transforming it into information and, ultimately, knowledge. A knowledge pipeline is the set of systems and processes through which raw data becomes actionable knowledge.

For example, supermarket managers need to track inventory and prices. They get data from various sources, including providers. Once inside the organization, this data makes it into spreadsheets and dashboards, where managers decide what to stock. Retail prices make it onto systems that tell consumers how much things cost. Managers track variations over time to learn about pricing patterns. Basically, you can’t run a supermarket without a functioning knowledge pipeline.

For knowledge workers, the pipeline is the job. These folks are responsible for gathering, compiling, synthesizing, transforming, communicating, and otherwise working on and with stuff moving through the pipeline. It’s a central part of every business, not just supermarkets. A hospital, for example, must coordinate schedules, patient data, diagnostics, and billing: all parts of a complex knowledge pipeline.

In a well-functioning business, the pipeline ingests the right data and transforms it into information that allows people to make good decisions in a timely manner. Information is processed and stored to become knowledge that allows the organization to improve over time.

Today, these transformations are done by people using the traditional tools of knowledge work: email, spreadsheets, dashboards, enterprise apps, databases, etc. They emerged in a world where only humans performed these transformations and communicated the resulting information.

Humans are well-suited for many jobs that entail working with information. For example, discussing performance evaluations with employees is best done by humans. But humans are less effective at other knowledge activities, such as making thousands of calculations in real time or spotting patterns in large data sets.

Today, bottlenecks are more likely to develop due to humans’ inability to process information at scale and in a timely manner than in technical limitations. AI can help. To answer the question of how to best use AI, managers must first understand their knowledge pipeline:

  • How is data coming in?
  • How is it processed?
  • Who needs to know what by when?
  • What information do we not have because we either can’t get it or can’t process it at scale?

AI can relieve bottlenecks so management can use information more effectively and efficiently. It can also unlock new ways of transforming data to information to knowledge. While doing this isn’t as sexy as having chatbots make better slides, it’s much more impactful.

Businesses stand to realize significant competitive gains by mapping their knowledge pipelines and adding AI agents to the flows. AI ROI doesn’t start with models, but by understanding how information flows in your business — and designing means for it to flow faster, clearer, and smarter than ever.