5 min read
AI isn’t simply changing how work gets done. It’s changing the economics, operating model and management of the modern enterprise.
The debate around AI often starts with the wrong question.
“Will AI replace people?”
I don’t believe that’s where the real transformation lies.
The more important question is this:
If AI can perform much of the routine work involved in serving a customer, how does that fundamentally change the economics of the enterprise?
Every major technology shift has changed how organisations operate. Steam power transformed manufacturing. Electricity enabled mass production. The internet redefined how businesses connected with customers.
AI is different again.
It isn’t simply another technology investment. It is creating an entirely new operating model.
Consider one of the most common customer journeys in financial services: applying for a mortgage.
Today, that journey involves mortgage advisers, customer service teams, underwriters, fraud specialists, compliance teams and administrators, all working across multiple systems. A customer experiences a single interaction, but behind the scenes there are numerous hand-offs, duplicated activities and manual decisions.
By the time organisations account for salaries, National Insurance, pensions, office space, management, software licences, compliance, quality assurance, training and operational overheads, that initial customer journey typically costs around £22–£25.
Now imagine designing that same organisation today, knowing AI exists.
Rather than one adviser carrying out every task, a digital workforce of specialist AI agents performs identity verification, document analysis, affordability assessments, fraud detection, product recommendations, document generation and customer communications simultaneously. Existing banking systems remain in place, but AI orchestrates the workflow across them.
To the customer it feels like one intelligent conversation.
Behind the scenes, dozens of specialist digital workers collaborate in seconds.
The remarkable part isn’t that this is now technically possible.
It’s that the economics are fundamentally different.
The Economics of AI
Running an AI-enabled mortgage journey is surprisingly inexpensive.
Using a modern enterprise AI platform, the runtime costs are measured in pennies. Even after including model inference, orchestration, monitoring and supporting infrastructure, the cost of executing the workflow remains extremely low. When enterprise platform licences, governance, cloud infrastructure and development costs are spread across millions of transactions, the economics become difficult to ignore.
| Cost per Mortgage Application | Traditional Model | AI-First Model |
|---|---|---|
| Labour | £14.00 | — |
| Operational Systems | £8.00 | £0.20 |
| AI Compute | — | £0.20 |
| Platform & Governance | — | £0.11 |
| Development (amortised) | — | £0.30–£0.80 |
| Total Cost | £22–£25 | £0.80–£1.30 |
This isn’t a marginal productivity improvement.
It’s a completely different economic model.
One of the first questions I hear whenever AI is discussed concerns energy consumption.
It’s a valid concern, but it’s important to separate training AI models from using them.
Training frontier models consumes enormous amounts of compute. Most organisations, however, will never train their own models. They’ll consume existing enterprise models through secure platforms.
A complete mortgage journey typically consumes around 0.02–0.15 kWh of electricity—less than 5 pence of power and approximately 5–40 grams of CO₂e. Compare that with the environmental impact of office buildings, employee commuting, heating, cooling, physical infrastructure and multiple operational hand-offs, and the discussion becomes far more balanced.
The environmental challenge is real, but so too is the opportunity to redesign how organisations consume resources.
AI Doesn’t Just Reduce Costs. It Redesigns the Operating Model.
This is where I think many organisations are still thinking too narrowly.
Most transformation programmes begin by asking:
“How do we automate today’s processes?”
A more interesting question is:
“If we were building this company from scratch today, knowing AI exists, what would it look like?”
Traditional organisations scale by hiring people.
More customers require more advisers, more administrators, more quality assurance teams, more managers and more office space.
Growth is directly linked to headcount.
An AI-first organisation scales very differently.
Routine operational work is carried out by a digital workforce, while people focus on activities requiring judgement, empathy, creativity and relationship management.
The operating model shifts from labour-intensive to intelligence-intensive.
| Traditional Enterprise | AI-First Enterprise |
|---|---|
| Growth through recruitment | Growth through digital labour |
| Large operational teams | Smaller, specialist expert teams |
| Sequential process hand-offs | AI-orchestrated workflows |
| Labour-intensive operations | Intelligence-intensive operations |
| Human processing | Human judgement and decision-making |
| Department-centric | Customer journey-centric |
Technology is no longer supporting the business.
Technology becomes the business.
This changes far more than operating costs.
- Margins improve.
- Service becomes available 24 hours a day.
- Customer journeys become dramatically faster.
- Innovation cycles accelerate.
- Organisations become significantly easier to scale.
The Next Management Discipline
I also believe AI will fundamentally change how organisations measure themselves.
For decades we’ve measured capacity through people.
- Headcount.
- Salary costs.
- Productivity.
- Employee engagement.
- Recruitment.
Over the next decade, every executive team will manage two workforces.
The first will be human.
The second will be digital.
Boards will increasingly ask questions such as:
- How many AI agents do we operate?
- What percentage of work is performed autonomously?
- What is our cost per digital worker?
- Where are AI agents underutilised?
- What return are we generating from digital labour?
In effect, organisations will develop a Digital Labour Balance Sheet alongside their traditional workforce planning.
Executive dashboards will evolve accordingly.
| Human Workforce | Digital Workforce |
|---|---|
| Headcount | AI Agent Count |
| Salary Costs | Compute Costs |
| Employee Productivity | AI Utilisation |
| Employee Engagement | AI Success Rate |
| Training Investment | Model Improvement Rate |
| Revenue per Employee | Revenue per Digital Worker |
This represents a profound shift.
For over a century, organisations have scaled by recruiting people.
Over the next decade, many will scale by expanding a managed digital workforce alongside their human one.
That requires new governance, new leadership capabilities and new measures of success.
My Perspective
I don’t believe AI is fundamentally about automation.
Nor do I believe it’s simply another technology programme.
I believe we’re witnessing the largest redesign of the enterprise since the arrival of the internet.
The organisations that succeed won’t necessarily have the biggest technology budgets or the largest AI models.
They’ll be the organisations that rethink how value is created.
They’ll redesign customer journeys rather than automate individual tasks.
They’ll manage digital labour with the same discipline they manage human talent.
And they’ll recognise that competitive advantage no longer comes from employing more people—it comes from combining human expertise with digital capability in ways that were simply impossible a few years ago.
The future isn’t AI replacing people.
It’s organisations built around two workforces—human and digital—working together to deliver better customer outcomes, operate at dramatically lower cost and create entirely new sources of competitive advantage.
That, in my view, is where the real transformation begins.