Where AI creates business value

In the second week of November, Andre Leclerc, operations manager for the eleven clinics of Cascade Ridge Veterinary Partners, was handed a job nobody had held before. The board had watched the autumn AI assessment with interest and some alarm. It now had a list of more than a dozen AI ideas, a handful of enthusiastic vendors, and no way to tell which ideas were worth money. It asked Andre to bring back an AI roadmap by the end of the first quarter of 2027: which ideas to fund, in what order, with what budget, and how the group would know whether they were working.

Andre was not a data scientist. He ran schedules, staffing, supplies and the daily problems of eleven busy clinics. That was exactly why the board chose him. "We have people who can build models," Lorraine Beaudry, the CFO, told him. "We need someone who can tell us which models are worth building." Andre decided to study for the AWS Certified AI Business Strategist exam (AIB-C01) along the way, and he planned to sit it in May 2027.

Noor Siddiqui, the data scientist running the group's AI assessment, agreed to coach him. Their Thursday-morning strategy sessions are the thread of this course. Noor brought the technical judgment; Lorraine joined for anything involving money; Dr. Simone Kaur, the medical director, joined whenever an idea came near clinical work; and Hannah Weiss, the privacy and security lead, joined whenever client data was involved, which was often.

This course and the exam

The AIB-C01 exam is aimed at business leaders, consultants, program managers and others who decide where AI is used and are accountable for its results. It does not test coding or hands-on AWS configuration. Its questions are scenario based, in two formats: multiple choice, with one correct answer out of four, and multiple response, with two or more correct answers out of five or more options. The exam guide divides the content into four domains:

Domain 2 has three tasks. The first is developing AI strategies that line up with business objectives: finding high-impact use cases, deciding whether to build, buy or partner, prioritizing initiatives, recognizing when AI is the wrong answer, and preparing for transitions. The second is measuring and demonstrating value: key performance indicators, baselines, return on investment, leading indicators and cost control. The third is positioning AI for competitive advantage: reading the competitive landscape, spotting business model opportunities, understanding which advantages last, and setting the right level of investment. Each lesson in this course takes up part of that list, and the last lesson pulls it together into Andre's roadmap.

Results are reported on a scaled score from 100 to 1,000, with 700 needed to pass. Scoring is compensatory: you pass on the overall score, not on each domain separately. Unanswered questions count as wrong and there is no penalty for guessing. The exam readiness course covers format, timing and strategy in detail.

Value starts with the business, not the technology

Noor began the first session by turning Andre's list upside down. The board's list was organized by technology: a chatbot, a forecasting model, an image tool, a summarizer. "Every one of these is a solution looking for a problem," she said. "Start from the other end. What does Cascade Ridge need to get better at? Then ask which of those needs AI can help with."

This is the most important habit in AI strategy. Business value is a measurable improvement in something the organization cares about: more revenue, lower cost, lower risk, better experience for clients or staff, or faster and better decisions. A technology that produces impressive demonstrations but does not move one of those measures has not created business value, however clever it is. A strategy therefore begins with business objectives and works toward capabilities, not the reverse.

Cascade Ridge's objectives for 2027 were already written down in the board's annual plan. Andre copied them onto the whiteboard:

• Grow revenue per clinic without adding exam rooms, mainly by filling more of the appointment slots the clinics already have.
• Keep the emergency hospital safe and well staffed overnight, where technician turnover had been high.
• Reduce the administrative load on veterinarians and technicians, which staff surveys named as the main cause of burnout.
• Keep client trust high, measured by review scores and the share of clients who return within eighteen months.
• Hold the growth in operating costs below the growth in revenue.

"Good," Noor said. "Now every AI idea has to point at one of those lines, or it doesn't get a slot on the roadmap."

The levers AI can pull

It helps to name the ways AI creates value, because each lever is measured differently. Most AI benefits fall under five broad levers.

Two cautions came up straight away. First, one idea can pull several levers at once. The no-show model helps revenue (filled slots) and experience (clients who need a reminder get one), and it may cut staff time spent on phoning. A good business case names the main lever and treats the others as additional benefits rather than adding them all up uncritically. Second, some levers are easier to measure than others. Cost reductions and revenue appear in the accounts; experience and decision quality often show up only indirectly. Lesson 5 deals with measuring both kinds.

Mapping capabilities to outcomes

The exam guide asks you to identify high-impact AI use cases across business functions and to map AI capabilities to specific business outcomes. It names customer operations, sales and marketing, research and development, and software development as examples. A capability is what a type of AI can do: classify, predict a number, find patterns, extract information from documents, understand and generate language, recognize images, or plan and act through tools as an agent. An outcome is the business result you want. Strategy work is largely the work of joining the two with a believable chain of cause and effect.

Noor drew the chain as a value tree. At the left is an objective, such as growing revenue per clinic. It splits into drivers: more appointments kept, and more value per visit. Each driver splits again, until a branch is specific enough that an action can move it: fewer no-shows, for example. Only at that point does an AI capability appear, such as predicting which bookings are likely to be missed so staff can send an extra reminder or offer the slot to a waitlist. Next to the capability goes the measure that will show whether it worked: no-shows per 100 appointments.

A value tree does three useful things. It forces the chain of cause and effect to be stated, so weak links are visible: if extra reminders do not actually change client behavior, the whole branch fails. It shows where AI is only one part of the answer, because the reminder process, the waitlist and the reception staff matter as much as the model. And it gives each initiative a measure before anyone builds anything, which is the start of the baseline work in lesson 5.

Competitive advantage, briefly

The third task in Domain 2 asks how AI can give an organization an advantage over its competitors. Lesson 11 treats this properly, but the main idea belongs here. Many AI tools are available to everyone: any clinic group can buy the same scheduling software with the same built-in assistant. Those tools may be worth having, but they rarely set an organization apart, because competitors can match them quickly. Advantage that lasts usually comes from things competitors cannot copy easily: proprietary data, such as Cascade Ridge's three years of appointment history and 95,000 pet records; processes redesigned around AI rather than AI bolted onto old processes; skills and habits built up through practice; and the trust of clients who have seen the organization use AI carefully.

Andre found this reassuring. "So we don't have to be first," he said. "We have to be good at using what we have." Noor agreed, with a warning: being too slow has a cost too, if competitors use AI to offer clients something Cascade Ridge cannot.

Value that is real, not apparent

Before ending the session, Lorraine raised the question that would return in almost every later lesson. Vendors had shown her impressive figures: hours saved per week, tickets deflected, productivity up. "How much of that is real money?" she asked.

The honest answer is: it depends what happens next. If an AI tool saves each technician twenty minutes a night, that time has value only if it is used for something that matters: more patient care, fewer overtime hours, less burnout and turnover. If the twenty minutes simply disappear into the shift, the saving is real for the technician but invisible in the accounts. Strategists distinguish realized value, which shows up in a measure the business tracks, from potential value, which would show up if the organization changed something to capture it. A good plan says how potential value will be turned into realized value, for example by reducing overtime budgets, or by tracking staff turnover as a target measure.

Noor's rule of thumb for the board was simple: a benefit counts in the business case if it has a named owner, a measure, a baseline and a plan for capturing it. Anything else goes in a separate column labeled "other benefits", where it can support a decision but not carry it.