Beyond Prompting: How AI Adoption Changes Business Operations

By Caroline Kennedy | Published September 18, 2026

 

Beyond Prompting: How AI Adoption Changes The Way Business Operates

For the past couple of years, much of the conversation about using AI in business has centred on prompting. How do you write a better prompt? Which model should you use? How do you get a better answer?

Prompting matters. I often tell leaders that prompting is briefing. The quality of what you ask, the context you provide and the clarity of the outcome you want will influence what you get back.

But after working with leadership teams and businesses on AI adoption, I think we’re reaching a much more interesting stage.

The bigger opportunity isn’t simply helping people become better at talking to AI. It’s working out how AI becomes part of the way the business actually operates.

That distinction matters because a business can have hundreds of employees using AI every day and still make very little meaningful change to how the organisation works.

I recently delivered an AI keynote and practical session for the leadership group at 1834 Hotels. The organisation had already invested in Microsoft Copilot and its leaders had started experimenting with AI. The question wasn’t whether they should use it. The more useful question was: what becomes possible next?

That is a relatively small group when nearly nine in ten respondents say their organisations already use AI regularly somewhere in the business. The gap tells us why AI adoption and AI ROI cannot be treated as the same thing.

That took the conversation beyond individual prompts and into the work itself.

What information does a general manager repeatedly need before making a decision? What preparation happens every day or every week? What information is scattered across reports, emails, meeting notes, customer feedback, financial information and operational systems? What work could AI prepare automatically before a person even needs to ask for it?

Those are very different questions from, “What should I type into Copilot?”

And they’re increasingly the questions I think CEOs and executive teams should be asking.

 

Context Matters More Than The Perfect Prompt

Businesses are full of context. It sits in emails, spreadsheets, reports, policies, customer reviews, meeting notes, financial information, operating procedures and, rather inconveniently, inside the heads of people who have worked there for 15 years.

When AI has access to the right approved organisational context, its usefulness changes considerably.

Imagine a hotel general manager preparing for tomorrow. Rather than opening multiple systems, searching through emails and trying to remember what happened in the last owners’ meeting, AI can help prepare a briefing using agreed sources such as the daily operating report, roster, customer reviews and relevant financial information.

The leader can then ask better questions of that information.

What needs my attention tomorrow? Where are we seeing a pattern? What has changed? Which issue keeps recurring? What decision am I avoiding because I don’t have enough information?

This is where AI begins moving from content generator to decision-support capability.

It also explains why businesses can spend considerable amounts on AI licences without seeing equivalent business value. Giving people access to intelligence isn’t the same as redesigning work around it.

 

What Changes When AI Agents Enter the Business?

The next shift is already happening.

Instead of a person remembering to open an AI tool and initiate the same task every Monday morning, an agent can increasingly be given a defined job to perform repeatedly.

It might review agreed information, identify changes, prepare a briefing, surface anomalies or assemble the evidence someone needs before making a decision.

The human hasn’t disappeared. The initiation of the work has.

That distinction is important.

The objective isn’t to hand every decision to an autonomous system. It’s to identify the repetitive preparation, analysis and information gathering surrounding human work, then determine which parts AI can perform reliably.

This creates a much more useful leadership question: What should AI prepare, and what should humans still judge? Schedule preparation, not judgement. 

This has become one of the principles I use when talking to leaders about AI: Schedule preparation, not judgement.

AI can increasingly prepare information before people need it. It can summarise, compare, analyse, identify patterns, generate options and flag exceptions.

But consequential decisions still require context, judgement and accountability.

A general manager can receive an AI-prepared operational briefing each morning. A CEO can have information synthesised before an executive meeting. A sales leader can have customer activity reviewed before a pipeline discussion. A manager can arrive at a coaching conversation with relevant information already assembled.

The technology prepares.

The person judges.

The leader remains accountable.

That distinction becomes increasingly important as AI systems become capable of doing more without being explicitly prompted every time.

 

The Real Opportunity May Be Management Capacity

One of the most interesting consequences of this shift has very little to do with technology.

It’s management capacity.

Think about how much leadership time is consumed by searching for information, preparing documents, consolidating reports, summarising meetings, writing routine communications and repeating administrative work. If AI absorbs some of that preparation, the value isn’t simply the number of hours saved.

The real question is what leaders do with the capacity they get back. A manager could spend more time coaching people. A CEO could spend more time thinking about the future of the business. A salesperson could spend more time with customers. A general manager could spend more time noticing what’s actually happening in the operation.

Of course, organisations could also save two hours with AI and immediately fill them with another meeting. Human ingenuity remains undefeated.

But that would miss the larger opportunity.

 

AI Transformation Requires Workflow Redesign

This is where I think many businesses will encounter the next constraint. They’ll get reasonably good at using AI while leaving the underlying work almost completely unchanged.

The old process remains. The old approvals remain. The old reporting remains. The old meetings remain. We simply insert AI somewhere in the middle and perform one part of the old process faster.

That can create productivity.

It doesn’t necessarily create transformation.

The larger opportunity is to examine the workflow itself.

If AI can prepare this information automatically, why does someone still compile the report?

If the system can identify the exception, why does a manager review every transaction?

If a recurring briefing can be prepared before the meeting, why are six executives spending the first 20 minutes establishing what happened?

If organisational knowledge can be made accessible when people need it, why are we relying on somebody remembering who knows the answer?

These are operating-model questions, not prompting questions.

And that’s why I believe the next stage of AI adoption will increasingly become a leadership and business transformation challenge.

 

Where Should a Business Start?

Not with the tool. Start with the constraint.

Find one recurring piece of work that consumes time, slows decisions, creates inconsistency or requires people to repeatedly assemble the same information.

Then map it.

What triggers the work? What information is required? What preparation happens repeatedly? Where is human judgement genuinely necessary? What could be automated or prepared by AI? What approvals need to remain? And most importantly, what business measure should improve if you change it?

Then test the redesigned workflow.

That might be a daily operational briefing, weekly sales review, customer-feedback analysis, executive meeting preparation, hiring workflow, financial review or another recurring process.

One useful workflow embedded properly is worth considerably more than 100 clever prompts saved in a document nobody opens again.

 

Moving From AI Productivity to Business Performance

I wrote recently about the shift I’m seeing from AI productivity towards AI business performance.

That distinction is becoming more important.

Productivity asks: How can AI help me do this task faster?

Business performance asks: Should this work happen this way at all?

The second question is harder. It requires leaders to understand the technology well enough to recognise what has become possible, while understanding their business well enough to know what is worth changing.

That is where my own interest in AI sits.

I spent much of my career as a CEO leading organisations through change and disruption, using technology and changing market conditions to rethink how businesses operated and competed. AI is a new technology, but the leadership challenge underneath it is familiar.

The organisations that create the greatest advantage from AI may not be those that become the best at prompting.

They may be the ones willing to reconsider how work gets done in the first place.

 

About Caroline Kennedy

Caroline Kennedy is an award-winning former CEO, CEO and executive coach, and keynote speaker who has led multinational businesses generating up to $250 million in annual revenue. She works with CEOs, executive teams and organisations on leadership, business performance, transformation and the practical application of AI, helping leaders understand what emerging technology makes possible and how to turn that possibility into meaningful business value. If you’re ready to become a more effective leader and create better conditions for your people to perform, find out more about Caroline’s Executive Coaching.