AIYING CaseLab #01
More AI. More Spending. But Where Is the Value?
A Current Challenge for Professionals and Institutions
Artificial intelligence has moved rapidly from experimentation into everyday professional work.
Organizations are buying AI tools. Employees are experimenting with copilots and assistants. AI training programs are expanding. Pilots are multiplying. Leaders are talking about agents, automation and AI-first operating models.
But beneath all this activity sits a much more important management question:
What measurable value are we actually creating?
That question forms the first challenge in AIYING CaseLab.
CaseLab is not designed around hypothetical AI trends alone. It starts with current challenges visible in the real world, asks professionals to exercise judgment, and then applies the AIYING approach to move from problem to accomplishment.
The CaseLab journey is:
Current Challenge → Judgment → Action → Value
And every applicable case ultimately produces an AIYING Value Record:
Challenge → Intervention → Before → After → Value → Human Judgment → Evidence
PART 1 — THE CHALLENGE
AI adoption is no longer the whole story
The evidence suggests that AI use itself is becoming increasingly normal.
The Thomson Reuters Future of Professionals Report 2026, based on more than 1,800 professionals across 62 countries, reports that 74% of professionals use AI several times a week, with 44% using it multiple times a day. (Thomson Reuters)
Yet widespread use does not automatically mean organizations are realizing widespread value.
Thomson Reuters identifies what it calls an AI value gap: organizations may have AI ambitions and employees may already be using AI, but strategy, tools, training, workflows and day-to-day execution are not necessarily aligned. Among professionals working in organizations with a named AI strategy, 35% say everyday practice does not match that strategy. (Thomson Reuters)
There is also an important difference between merely having AI available and deliberately organizing around value creation.
In organizations with a named AI strategy, 66% of professionals say AI is meeting or exceeding expectations for creating value at work. Where there is no active strategy, that figure falls to 22%. (Thomson Reuters)
McKinsey's June 2026 operational-excellence research points to a similar scaling problem: almost 90% of organizations surveyed said they were at least experimenting with AI, while only 7% reported scaling it across the enterprise. (McKinsey & Company)
And McKinsey's August 2026 State of AI report is tellingly subtitled “On the road to ROI”, reflecting how the conversation has moved from simple adoption toward capturing organizational returns. (McKinsey & Company)
Recent reporting adds another dimension. The Wall Street Journal reported in September that a Gartner survey of roughly 1,300 corporate leaders found 85% of functional leaders planned to increase AI spending in 2026, while 23% said they did not know their AI ROI. (The Wall Street Journal)
Taken together, these findings point toward an important contemporary management challenge:
AI activity is growing faster than many organizations' ability to demonstrate AI value.
THE CASE
Consider an organization that has spent the past year actively encouraging AI adoption.
It has:
purchased AI licenses;
conducted employee training;
encouraged experimentation;
launched several pilots;
established internal AI initiatives;
and seen increasing employee usage.
Management is pleased with the apparent momentum.
The leadership team therefore begins discussing a larger AI budget for the coming year.
During the meeting, however, the CFO asks:
“What measurable value have we actually created?”
The team begins presenting evidence.
There are hundreds of AI users.
Thousands of prompts have been submitted.
Training participation is high.
Several departments have conducted pilots.
Employees can demonstrate impressive AI-generated reports, presentations, analyses and summaries.
But the CFO asks again:
“What became better because of AI?”
Now the conversation becomes more difficult.
Can the organization demonstrate that:
cost decreased?
cycle time improved?
quality increased?
errors decreased?
customers received better service?
employees gained productive capacity?
decisions became faster or better?
new revenue or capability was created?
The organization has considerable evidence of AI activity.
It has much less evidence of AI accomplishment.
THE DECISION
Imagine that you are advising the leadership team.
What should it do next?
A. Buy more powerful AI tools
Perhaps existing tools simply aren't capable enough.
B. Conduct more AI training
Perhaps employees need better prompting and AI skills.
C. Require everyone to use AI
Perhaps adoption needs to become mandatory.
D. Identify specific workflows where AI can create measurable value
Instead of starting with technology adoption, start with business outcomes.
E. Something else
What would you recommend?
Before reading further, make your decision.
Because CaseLab is not primarily about finding the answer supplied by AI.
It is about developing professional judgment in the presence of AI.
PART 2 — THE AIYING SOLUTION
Our judgment: Start with D — but go further
Identify specific workflows where AI can create measurable value.
The crucial change is the starting question.
Many AI initiatives effectively begin with:
Where can we use AI?
AIYING proposes a different starting point:
Where do we need to create value?
This distinction matters.
If AI usage becomes the objective, organizations naturally measure licenses, adoption, prompts, pilots and training.
If value creation becomes the objective, management starts measuring outcomes.
Apply the AIYING Lens
AIYING uses TAJ — Think → Ask → Judge to structure human-AI collaboration.
Let's apply it to the case.
1. THINK — Start with the outcome
Suppose one recurring organizational problem is slow customer-complaint resolution.
The conventional AI question might be:
How can we use AI in customer service?
That question is too broad.
AIYING begins by clarifying the desired accomplishment:
Reduce complaint-resolution time while maintaining or improving response quality and appropriate human accountability.
Now AI has a purpose.
We have moved from:
Technology → Possible uses
to:
Challenge → Desired outcome → Possible AI intervention
2. Map the existing work
Before introducing AI, understand how work currently happens.
A simplified complaint workflow might look like:
Complaint → Read → Classify → Route → Investigate → Decide → Respond → Follow up
Now examine each stage.
Where is repetitive effort occurring?
Where is information being interpreted?
Where are patterns important?
Where are delays occurring?
Where does professional judgment matter?
Where does accountability require a human decision?
Only after understanding the work should we decide where AI belongs.
3. ASK — Use AI to explore the intervention
Instead of immediately asking AI to automate the workflow, use it first as an analytical partner.
AIYING Power Prompt
Analyse this customer-complaint workflow. Identify steps where AI could reduce repetitive effort, detect patterns, improve prioritisation or assist decision-making. For each opportunity, explain the expected value, risks, required human judgment and how the improvement could be measured. Do not assume that every step should be automated.
AI might identify opportunities such as:
classification
AI could categorize incoming complaints.
summarisation
AI could turn lengthy submissions into concise issue summaries.
urgency detection
AI could flag potentially serious complaints for faster attention.
pattern detection
AI could identify recurring issues across large volumes of complaints.
response assistance
AI could prepare an initial response draft for human review.
But AIYING does not stop with what AI can do.
That leads to the most important TAJ step.
4. JUDGE — Should AI do it?
AI capability is not the same as organizational suitability.
For every proposed intervention, ask:
Can AI do this?
and then:
Should AI do this?
and finally:
Will doing this create enough value to justify the change?
Evaluate each intervention against factors such as:
Accuracy | Risk | Cost | Privacy | Reliability | Human oversight | Value potential
This preserves a crucial principle of AIYING:
AI contributes capability. Humans retain judgment and accountability where they matter.
This is particularly important in professional environments. Thomson Reuters' 2026 report explicitly emphasizes that while AI can act as a powerful force multiplier, professional judgment, relationships and accountability remain human responsibilities. (Thomson Reuters)
5. INTERVENE — Redesign one workflow
Do not immediately attempt enterprise-wide transformation.
Start with a workflow where the challenge and outcome are sufficiently clear to measure.
Before
Complaint
↓
Manual reading
↓
Manual classification
↓
Routing
↓
Investigation
↓
Response
AI-assisted pattern
Complaint
↓
AI-assisted classification + summarisation
↓
Human validation
↓
Priority routing
↓
Investigation
↓
AI-assisted response draft
↓
Human approval
Notice what has happened.
AI has not simply been “added.”
The work pattern has been redesigned.
And human judgment has not disappeared.
It has been deliberately positioned where it adds the most value.
6. ADD VALUE — Ask what became better
This is where many AI initiatives should become much more disciplined.
Don't measure only:
How many employees used AI?
Measure:
What became better because AI was introduced?
Possible value dimensions include:
Time ↓
Did the workflow become faster?
Cost ↓
Did the cost per transaction decrease?
Repetitive effort ↓
Did professionals recover capacity for higher-value work?
Errors ↓
Did accuracy or consistency improve?
Quality ↑
Did the resulting work improve?
Customer experience ↑
Did customers receive faster or better service?
Capability ↑
Can the organization now accomplish something it previously could not?
Opportunity ↑
Did AI enable new services, products or revenue possibilities?
This is the transition from AI use to AI value.
7. Establish the BEFORE
Before deploying the intervention, capture baseline measurements.
For our complaint example, the organization might measure:
average handling time;
average response time;
first-routing accuracy;
number of complaints requiring rerouting;
customer satisfaction;
employee effort per complaint;
escalation rate;
and quality/compliance indicators.
Without a baseline, improvement becomes difficult to prove.
This is why measurement should begin before implementation rather than being added after a successful-looking pilot.
8. Establish the AFTER
Run a controlled pilot.
Measure exactly the same indicators.
Suppose the pilot produces:
Average handling time:
20 minutes → 12 minutes
First-routing accuracy:
82% → 94%
Average response time:
18 hours → 9 hours
These figures are illustrative teaching numbers for this CaseLab case. They are not results reported by Thomson Reuters, McKinsey, Gartner or another external study.
That distinction is important.
CaseLab will separate verified external evidence from illustrative case assumptions.
Now compare the before and after results.
Did the improvement actually occur?
Was quality maintained?
Were there unintended consequences?
Were employees merely processing more work, or was released capacity redirected toward higher-value activities?
Did customers experience the improvement?
Only then can the organization reasonably claim an accomplishment.
9. PROVE — Create the AIYING Value Record
Every applicable AIYING CaseLab concludes with a simple artifact.
AIYING VALUE RECORD
Challenge
Customer complaints take too long to process.
Intervention
AI-assisted classification, summarisation, prioritisation and response drafting, with human validation and approval.
Before
Record baseline handling time, response time, routing accuracy, quality and other relevant indicators.
After
Record the same indicators following the controlled pilot.
Value
Determine the demonstrated improvement in time, effort, quality, customer experience, capacity or other outcomes.
Human Judgment
Humans validate classification, investigate substantive issues, handle exceptions and retain responsibility for consequential decisions and final responses.
Evidence
Before-and-after operational data, quality reviews, customer indicators and pilot records.
That gives us:
Challenge → Intervention → Before → After → Value → Human Judgment → Evidence
This is no longer simply an AI experiment.
It is a documented AI accomplishment.
10. SCALE — But scale the right thing
Only now should leadership ask:
Can we scale this?
And even here, there is an important distinction.
Don't scale AI merely because a pilot successfully demonstrated that the technology works.
Scale because the intervention demonstrated repeatable value.
A useful progression becomes:
Prove one workflow → Learn → Improve → Standardize → Scale
This also aligns with the broader 2026 evidence that the difficult part of organizational AI is increasingly not experimentation itself but translating AI into operational practice and scalable value. McKinsey reports widespread experimentation alongside far more limited enterprise scaling, while Thomson Reuters finds significant gaps between AI strategy and professionals' everyday work. (McKinsey & Company)
What was wrong with the original management question?
Nothing.
The CFO was asking exactly the question that needed to be asked:
“What measurable value have we actually created?”
The problem was when it was being asked.
It appeared after tools had been purchased.
After training.
After experimentation.
After pilots.
After adoption.
AIYING suggests bringing the question to the beginning.
Before every substantial AI initiative, ask:
What value are we trying to create, and how will we know whether we created it?
That one question changes the nature of AI adoption.
From AI Adoption to AI Accomplishment
Organizations can think about AI maturity through a simple distinction.
AI Activity
Tools → Licenses → Training → Prompts → Pilots → Usage
These indicators tell us whether AI is being adopted.
They remain useful.
But they do not necessarily tell us whether value has been created.
AI Accomplishment
Challenge → Outcome → AI Intervention → Human Judgment → Value → Evidence
These indicators tell us whether AI helped accomplish something worthwhile.
Organizations need both.
But they should never confuse the first with the second.
The AIYING Management Takeaway
The objective is not to maximize the amount of AI inside an organization.
The objective is to improve the organization's ability to accomplish valuable outcomes.
That gives CaseLab #01 its central principle:
Don't scale AI usage. Scale proven AI value.
And perhaps an even more fundamental question for every AI initiative:
Don't ask only: “Where can we use AI?”
Ask: “Where can we create value—and how will we prove it?”
That is the shift from using AI to Accomplishing with AI.
About AIYING CaseLab
AIYING CaseLab explores current challenges faced by professionals and institutions and examines them through a practical AIYING lens.
Each case begins with a real contemporary signal.
Readers are first invited to exercise their own judgment.
Then the case moves through:
Current Challenge → Judgment → Action → Value
Where applicable, it concludes with an:
AIYING Value Record
Challenge → Intervention → Before → After → Value → Human Judgment → Evidence
The purpose is not merely to learn more about AI.
It is to learn how to accomplish something valuable with AI—and prove it.
AIYING — Accomplishing with AI
Sources used and verified
The principal evidence for this case comes from the Thomson Reuters Future of Professionals Report 2026, based on 1,816 professionals across 62 countries; its report and accompanying analysis support the adoption, strategy-execution, value and professional-judgment claims used above. (Thomson Reuters)
The scaling context is independently supported by McKinsey's June 2026 operational-excellence survey and its August 2026 State of AI research. (McKinsey & Company) Recent Wall Street Journal reporting on Gartner's corporate-leader survey provides additional evidence of the tension between increasing AI spending and uncertain ROI. (The Wall Street Journal)
Fact-check note: All externally attributed percentages above are drawn from the cited sources. The complaint-management scenario is a generalized teaching case, and the 20→12 minutes, 82%→94%, and 18→9 hours figures are explicitly illustrative—not presented as external research findings.
No comments:
Post a Comment