McKinsey’s 26 August 2026 article, “The decision dividend: How AI creates economic value,” caught my attention because it describes, in corporate terms, something very close to the shift we have been exploring in financial planning. Its central argument is that AI’s biggest economic gains may not come from replacing labour, but from making better decisions faster, more frequently and at dramatically lower cost. For us, that raises a more human question. If AI is reducing the cost of intelligence, how much of that benefit is being passed on to the person making the financial decision? And should we begin measuring not just decision throughput inside institutions, but agency throughput — the extent to which people become more able to understand, choose and act for themselves?
Not “did we give good advice?”, but “did the person leave more capable than they arrived?”
For most of modern financial services, good financial decision-making has been expensive.
Not because thinking itself is inherently expensive.
Because access to information, analysis, modelling, expertise and judgement has traditionally required people.
People gather the data.
People interpret it.
People compare alternatives.
People calculate consequences.
People explain the options.
People help make the decision.
That made financial intelligence scarce.
And when something is scarce, somebody gets to control access to it.
In financial services, that access has usually been bundled into an intermediary relationship.
You want better financial decisions?
You employ an adviser.
You pay an ongoing fee.
You place assets under management.
You buy the product.
You enter the institution.
That model made economic sense when intelligence was expensive.
But something fundamental has changed.
AI is making intelligence cheap.
And financial services has not yet passed the saving to the customer.
The decision dividend
McKinsey recently described what it calls the “decision dividend.”
Its argument is important.
The greatest economic value from AI may not come from replacing workers.
It may come from enabling organisations to make decisions faster, more frequently and at dramatically lower cost.
AI can analyse information.
Model alternatives.
Identify patterns.
Test scenarios.
Challenge assumptions.
Surface risks.
Compare options.
Explain consequences.
And it can do much of this at a marginal cost approaching zero.
The significance is difficult to overstate.
For centuries, human intelligence has been economically scarce.
AI is beginning to change the cost curve of intelligence itself.
McKinsey cites research suggesting that some decision-making tasks that previously cost tens of dollars in human labour can now be performed for fractions of a cent.
That creates an enormous economic dividend.
But dividends have recipients.
So the question is not simply:
How much value will AI create?
The more interesting question is:
Who gets the value?
Financial services currently has an obvious answer
Look at how AI is mostly being deployed across financial services.
AI helps advisers write suitability reports.
AI summarises meetings.
AI prepares client communications.
AI analyses portfolios.
AI automates administration.
AI helps firms serve more customers with fewer staff.
AI allows an adviser to handle more clients.
AI reduces operating costs.
All perfectly rational.
But notice who captures most of the economic benefit.
The institution.
The intermediary.
The shareholder.
The adviser.
The customer may receive a slightly faster service.
But the commercial architecture often remains unchanged.
A customer who previously paid 1% of their assets each year may still pay 1%.
An adviser who previously served 100 clients might now serve 200.
The cost of producing the intelligence falls.
The price of accessing the relationship does not necessarily fall with it.
That should make us pause.
If technology reduces the cost of making a financial decision by 90%, why should the customer continue paying roughly the same price for access to financial intelligence?
We have seen this before
Technology frequently creates enormous productivity gains before those gains are distributed.
The first beneficiary is usually the producer.
A manufacturer automates its factory.
Costs fall.
Margins rise.
Eventually competition forces some of those savings towards consumers.
Prices fall.
Quality rises.
New entrants appear.
Entire markets are restructured.
Financial services may now be approaching the same point.
Except there is a complication.
Much of financial services is not priced according to the cost of producing advice.
It is priced according to the assets or products attached to the customer.
That disconnect matters enormously.
Imagine two people requiring exactly the same intellectual work.
One has £100,000.
The other has £1 million.
If both pay 1% of assets annually, one pays £1,000 and the other £10,000.
The cost of the underlying intelligence did not increase tenfold.
The price did.
That was already difficult to justify.
AI makes the question harder.
Because the cost of producing much of the analysis is now collapsing.
The old model was built around scarce expertise
Traditional financial advice contains an important hidden assumption:
expertise is scarce.
If expertise is scarce, the rational response is to organise distribution around experts.
The adviser becomes the gateway.
The client brings the problem.
The adviser interprets it.
The adviser produces the recommendation.
The adviser often implements it.
The adviser monitors it.
The client therefore becomes dependent upon continued access to the adviser.
That relationship can be valuable.
Sometimes extremely valuable.
But scarcity shaped the architecture.
AI changes the scarcity assumption.
Financial intelligence can increasingly become available directly to the individual.
Not perfect intelligence.
Not omniscient intelligence.
Not necessarily intelligence capable of replacing specialist human judgement in every circumstance.
But enough intelligence to transform what ordinary people can understand and decide for themselves.
That changes the architecture of financial planning.
From AI-enabled advice to AI-native agency
There are two very different ways financial services can respond.
The first is AI-enabled advice.
The existing system remains largely intact.
The adviser remains the centre of the architecture.
AI helps the adviser work faster.
More clients can be processed.
Administration becomes cheaper.
Advice production becomes more scalable.
This may improve efficiency.
But fundamentally it preserves the existing relationship.
The second possibility is much more interesting.
AI-native agency.
Instead of asking:
How can AI help advisers advise more people?
We ask:
How can AI help more people understand and make financial decisions themselves?
That produces a completely different system.
The individual becomes the centre.
AI helps them organise their information.
Understand their circumstances.
Explore alternatives.
Model consequences.
Identify uncertainties.
Generate questions.
Recognise when expertise is required.
Human experts still matter.
Perhaps more than ever.
But their role changes.
They become an additional layer of intelligence rather than the permanent gateway to it.
The model becomes:
continuous agency with episodic expertise.
Rather than:
continuous expertise with episodic agency.
That is a profound difference.
Some decisions genuinely need an expert
None of this means people should make every financial decision alone.
There are moments when specialist expertise is extraordinarily valuable.
Complex pension decisions.
Tax planning.
Trusts.
Business structures.
Divorce.
Inheritance.
Financial exploitation.
Regulatory disputes.
Major investment decisions.
Irreversible transactions.
Sometimes the smartest decision a person can make is to bring in somebody who has spent twenty years learning something they have encountered for the first time.
But that does not automatically imply a permanent advisory relationship.
A three-hour problem does not necessarily require a thirty-year commercial relationship.
AI makes episodic expertise far more practical because the individual can increasingly perform the surrounding work themselves.
Gather the information.
Understand the terminology.
Explore the obvious options.
Prepare the questions.
Record the evidence.
Then bring in the expert precisely where expertise adds value.
The expert becomes more valuable because less expert time is wasted doing work that no longer requires scarce human intelligence.
Decision throughput is not enough
McKinsey suggests organisations should measure something called decision throughput.
Rather than counting AI licences, pilots or tools, organisations should ask how many decisions are actually being improved, accelerated or automated.
That is a useful measure.
But from the citizen’s perspective, I think we need another one.
Agency throughput.
How many more decisions can a person:
understand,
evaluate,
make,
and act upon
without unnecessarily surrendering control to somebody else?
That is a very different objective.
An institution might dramatically increase decision throughput while simultaneously reducing individual agency.
Imagine an AI system that decides which financial product should be promoted to you.
The organisation has improved decision throughput.
The institution is smarter.
But you are not necessarily more capable.
Now imagine an AI system that helps you understand your options, model different futures, identify conflicts of interest and decide whether you need professional assistance.
That increases something else.
Your capacity to act.
Your agency.
The distinction matters.
There are two possible AI revolutions in financial services
One makes institutions much more powerful.
The other makes individuals much more capable.
They are not the same thing.
AI could allow financial institutions to know more about customers than customers know about themselves.
Their income.
Their expenditure.
Their behaviour.
Their vulnerabilities.
Their preferences.
Their likely decisions.
Their propensity to buy.
Combine that with powerful predictive systems and AI could become the most sophisticated product-distribution technology financial services has ever created.
But exactly the same technology could be placed on the other side of the table.
Imagine every person having their own financial intelligence system.
One that understands their goals.
Their resources.
Their commitments.
Their family.
Their health.
Their work.
Their values.
Their future possibilities.
One that can interrogate financial products.
Challenge assumptions.
Compare providers.
Model consequences.
Read documents.
Prepare questions.
Spot inconsistencies.
And remember everything.
That changes the balance of power.
The question is not whether AI will transform financial services.
It will.
The question is:
whose intelligence will it augment?
The customer should receive the saving
If AI reduces the cost of financial intelligence, there should eventually be a consumer dividend.
Financial understanding should become cheaper.
Second opinions should become cheaper.
Scenario modelling should become cheaper.
Financial education should become cheaper.
Decision support should become cheaper.
Planning should become accessible to people who were previously uneconomic for traditional advice firms to serve.
And human expertise should become something people purchase because they genuinely need expertise — not because expertise is the only gateway to understanding.
This could expand the financial planning market enormously.
Today millions of people receive neither financial advice nor meaningful financial planning.
They fall into what the industry often calls the “advice gap”.
But perhaps we have misdiagnosed the problem.
Maybe millions of people do not need continuous financial advice.
Maybe they need continuous access to financial intelligence.
Those are not the same thing.
AI suddenly makes the second possibility economically viable.
The advice gap may really be an intelligence gap
For years the industry has asked:
How do we provide advice more cheaply?
That question already assumes advice is the product.
A better question might be:
What capability is the person actually missing?
Sometimes they need expertise.
Sometimes reassurance.
Sometimes calculation.
Sometimes information.
Sometimes modelling.
Sometimes challenge.
Sometimes another pair of eyes.
Sometimes they simply need help organising their thoughts.
Historically those functions were bundled together inside something called “financial advice” because humans were required to provide all of them.
AI allows those functions to be separated.
That could be one of the most important consequences of the technology.
We may discover that what people need is not necessarily an adviser.
They need access to intelligence.
And occasionally they need an expert.
This changes what a financial planner can become
This should not be seen as an attack on financial planners.
It could create a better profession.
When machines perform more routine analysis, humans can concentrate on the things humans are unusually good at.
Judgement.
Context.
Empathy.
Experience.
Challenge.
Wisdom.
Holding difficult conversations.
Recognising what the numbers cannot see.
Helping somebody confront uncertainty.
Helping somebody discover what they actually want.
The planner stops being the owner of the intelligence.
They become part of the person’s intelligence network.
A second brain when another brain is useful.
That seems to me a much healthier professional relationship.
Because the purpose of expertise should not be to create dependency.
It should be to increase capability.
So who gets the decision dividend?
That may become one of the defining questions of the AI economy.
If the cost of intelligence collapses while customers continue paying the same intermediary charges, institutions capture the dividend.
If AI simply allows advisers to serve twice as many clients at the same percentage fee, intermediaries capture the dividend.
If AI becomes a better engine for distributing financial products, manufacturers capture the dividend.
But if people gain cheap access to sophisticated financial understanding, individuals capture it.
That is the opportunity.
We can build an AI-enabled version of the existing financial system.
Or we can build something different.
A system in which every person has access to powerful financial intelligence.
A system where expertise arrives when expertise is genuinely required.
A system where technology increases people’s capacity to understand, choose and act.
The economics of intelligence are changing.
The architecture of financial services eventually will too.
The question is whether the savings produced by AI remain inside the institutions that deploy it —
or whether some of that extraordinary decision dividend finally reaches the person making the decision.
Because the greatest social dividend from AI may not come from giving institutions better intelligence.
It may come from giving individuals greater agency.
Academy of Life Planning 