If AI Is So Powerful, Why Are We Still Optimising Expense Reports?
Lessons from Davos: Most organisations say they are investing in AI. Most are still optimising expense reports. That gap is no longer harmless - it’s now strategic risk.
During World Economic Forum in Davos, we ran the REWIRE program with a single mission: move beyond AI theater and into execution reality.
The shift we wanted to see? Concepts to practice.
As I said at during the opening:
"In 2025, Davos AI sessions were all about policy frameworks and governance - giving leaders an incredibly good excuse to do nothing. This year? We want to see people practically and actively using AI and getting extreme levels of value."
What became crystal clear across every session: the companies succeeding aren't the ones with the best AI tools. They're the ones connecting efficiency gains to revenue growth.
Most get neither.
And the gap showed up in the smallest moments.
The Imagination Gap
One question revealed everything:
"What forces will compel your organization to pursue AI-driven growth and efficiency this year?"
The idea wasn’t to identify nice-to-have improvements - it was to uncover what must be addressed if an organisation wants to produce real impact this year.
But many of the responses landed in familiar territory:
speeding up expense reports
automating administrative tasks
shaving minutes off writing emails
Useful - but not market-moving.
The real distinction is simple: efficient tasks vs strategic moves.
Why does this pattern repeat? Three forces: departments optimize what they control, vendors sell point solutions, and time-savings ROI is easier to measure than revenue growth.
Result: death by a thousand productivity improvements. Local optimisation is the enemy of strategic impact.
If the things you point to as the biggest potential for AI are internal frictions, you are optimising the status quo instead of challenging it. Many AI portfolios today are defence budgets for the current operating model - not investments in the next one.
What Real Value Capture Looks Like
To see the difference between tactical optimisation and strategic value capture, consider these examples from real organisations and patterns observed:
The $1B Model That Delivered $0
Dr. Eva-Maria Hempe from INVIDIA shared the cautionary tale we heard repeatedly:
A global retailer introduced a well-funded AI-based demand forecasting model with massive accuracy increases - 4% of company revenues in estimated benefits. But the reality is: it drove no measurable impact because it never touched the core decision loops of the business.
The AI worked. The efficiency gains were real. When decisions don’t change, results don’t change. But Dr. Eva-Maria Hempe from INVIDIA while the AI predicted demand perfectly procurement still ordered on last year's schedule, ie. the organization didn't change.
This isn’t a technology failure - it’s a strategic framing failure. When AI doesn’t influence the decisions that shape customer value, revenue, or cost structure, it simply becomes a cost with no return.
That’s expense-report thinking at enterprise scale.
2. News Agencies: Amplifying Strategic Advantage
Prof. Vijay Gurbaxani from UC Irvine shared the counter-example:
Rather than using AI to cut costs alone, a global news agency used it to reinforce their strategic advantage - delivering more original reporting faster than competitors.
The approach:
Efficiency: AI writes faster, lower cost.
Strategy: What makes the news agency unique? Original coverage.
Reinvestment: Deploy savings to underserved markets.
Growth: Stronger competitive position, premium pricing. Speed and originality became measurable competitive metrics.
They didn't just automate.
The productivity gains didn’t get pocketed. They were reinvested into the growth engine of the business. Productivity gains only matter when they are redeployed, not absorbed.
That's the opposite of expense report thinking. That's business model innovation.
3. Our Own Version: 31,000 Minutes of Capacity Unlocked
On a practical level, one of the examples came from our own business:
By automating the transcription, synthesis, and CRM integration of over 31,000 minutes of meetings in 2025, we didn’t just save time - we reduced decision preparation cycles and increased execution speed across projects to achieve measurable business impact.
That’s not “automated note-taking.” That’s capacity creation.
That’s where real organisational leverage lies.
The Hidden Cost Trap Nobody's Talking About
While we are all enthusiastic about introducing AI to our organizations, one important fact needs to be made visible.
"The new models take up 40 times the compute tokens that models did a year ago."
Costi Perricos from Deloitte mentioned this as an insights from broader industry research. Deloitte has publicly noted that demand for compute capacity is increasing much faster than hardware efficiency gains - with forecasted growth in AI workload demands of roughly 4–5× per year out to 2030.
40x. Not 40%. Times.
What this means:
If you let every department pick their own AI tools, you just committed to 40x cost increases across fragmented infrastructure.
The expense report approach: Scattered tools → Costs explode → CFO panics → AI budget cut → Zero ROI
The news agency approach: Unified platform → Economies of scale → Can afford to scale both efficiency AND growth
This is why we help organisations design AI-native workflows before scaling tools.
Infrastructure decisions aren't technical. They're strategic.
Fragmented AI stacks are the new shadow IT - just more expensive.
The Monday Test — A Simple Litmus for Strategy
One of the most actionable Davos litmus tests is what Mark Turrell and I call the Monday Test:
"What will you do differently on Monday?"
Not what you learned. Not what inspired you. What measurable action changes this week?
If the answer is:
“deploy more tools”
“run more pilots”
“evaluate another model”
…then you are still on the tactical treadmill.
A real AI strategy produces answers like:
“redirect savings into new product initiatives”
“reduce time to revenue by X% this quarter”
“shift headcount from routine tasks to growth areas”
“connect AI insights directly into decision workflows”
The Monday Test separates AI strategy from discussion decks.
In other words, ask yourselves 3 questions to ensure we focus on the right initiatives:
Revenue Test: Does this AI application directly influence what customers pay for?
Decision Test: Does it change how we make decisions that shape competitive position?
Reinvestment Test: Can efficiency gains fund growth in our strategic advantage areas?
If no to all three → You're optimizing expense reports at scale.
Scale doesn’t fix weak focus. It multiplies it.
So Where Do You Actually Stand?
The patterns described here are not theoretical. We’ve translated them into a simple diagnostic that shows whether AI is changing how decisions are made — or just optimising the edges.
If you want to see it for yourself, you can run it here:
Click on the picture above to access the assessment
It takes 5 minutes and highlights where structural constraints limit impact.
The Real Challenge
Prof. Vijay Gurbaxani from UC Irvine framed the stakes:
"We are at the dawn of the next industrial revolution. It's not just individuals and companies that will succeed and fail. It is also countries and societies."
$7 trillion being invested in AI infrastructure by 2030.
So yes, automate your expense reports.
But if that's where your strategic imagination ends, don't be surprised when you capture neither the efficiency gains nor the revenue growth you were promised. AI strategies that start with tools instead of value almost always underperform.
The companies winning aren't choosing between efficiency and growth.
They're connecting the dots to unlock both.
Monday Action: Run Your AI Portfolio Audit
Pull your list of AI initiatives. For each one, answer:
What business decision does this change?
What revenue or strategic advantage does it create?
Where do the efficiency savings get reinvested?
Can't answer all three? Kill it or redesign it. That's the difference between transformation theater and competitive advantage.
Or run an AI-Native Readiness Check to see where execution may be constrained.
Need help building what comes after the assessment? See how we work.
Working in pharma, medtech, or health insurance? See REWIRE for Health.
