The AI Playbook Top Companies Use

I'm Katie Drummond, Global Editorial Director at Wired,

and I'm sitting down today with Dan Swan,

a global co-leader within McKinsey & Company's

Tech and AI practice.

A few years ago, McKinsey's research produced Rewired,

a playbook for how companies actually transform

in the age of digital and AI.

That thinking has only gotten sharper

and more urgent since.

Most companies aren't actually transforming.

They're buying software

and calling it a transformation.

The ones getting it right

are doing something fundamentally different.

They're rewiring how their people,

their workflows, and their leadership operate.

That research draws on years of McKinsey client work

with companies like Toyota and DBS.

Dan advises clients globally,

and he's here to walk us through what's actually working,

what isn't, and what most leaders are still getting wrong.

Dan, thank you for being here.

I'm delighted to have you.

I'm excited to be here.

Now, we always start these conversations

with a little bit of rapid fire,

sort of like a warmup if you're game.

Sounds good.

All right.

Well, what is the most active text thread

you are on right now?

I would probably say the one with my family.

I'm one of four.

My parents have 11 grandkids.

No. Yeah.

So with 21 people,

there's always something going on that's interesting.

That's a lot of people

on one text thread. It is.

What is a piece of technology that changed your life?

I was an early adopter of the fitness wearables.

So I'm now, without oversharing,

about seven and a half years into taking

at least 10,000 steps every single day.

And then the last couple years I've adopted, you know,

learning about my sleep patterns

and the strain and recovery.

And so it combines two of my favorite things,

being active and data.

What does the algorithm know about you?

I guess other than maybe some of your sleep

and fitness habits.

I think more broadly, just kind of I have a common...

My engine's always going, right?

So I don't sit much,

but I also just the way my brain works.

I love jumping from one thing to the next.

I want one short-term prediction from you.

In the next three to six months,

what's something you think we're all going to see?

We've kind of gone through this moment of people going

from AI is interesting

to now we've got to scale and it's really important.

I think we're going to turn the next page

over the next three to six months.

It'll be all about really how you think

about the ROI of AI.

People are more and more focused on the return right now.

I think the investment portion will be super interesting.

Right?

What's the cost of a token

versus the benefit you get

from investing in that token,

I think is something that not a lot of folks

have thought about yet.

I think it'll be a real game changer.

What's a piece of tech you wish existed

but doesn't yet?

This is a dangerous one.

We've so democratized innovation of technology

that anything I think of,

I'm sure someone out there is thinking of it

and doing it.

But I'll kind of stay in the wellness arena.

I have tried multiple times to be disciplined enough

to track my nutrition and hydration,

and I've never been able to stick with it.

So if there was a way to automatically track nutrition

without totally going into the matrix,

I'd be game.

That's super interesting.

I mean, there will be an AI solve for that for sure.

I'll be an early adopter.

For sure, I believe it.

Let's talk about Rewired. Sure.

Why this book and why now?

Well, the interesting, you know,

the first version of Rewired

was all about digital transformation.

And this is all about how companies are adopting

and seeing success from AI.

So I think just that adjustment and that change

has been a really important topic

and the timing for given how prevalent it is.

I think the other thing

that's been really interesting about it

is some of the things we learned

about driving a successful digital transformation

are also true for AI transformation.

It's more than just the technology.

It requires the people.

So I think separating the facts from the hyperbole

is really important.

Now I'm curious, as the book was coming together,

you were looking at a lot of companies, right,

who are seeing real competitive advantages from AI.

What are they getting right?

What are they maybe doing differently

that others could learn from?

One is just this notion of focus, right?

Finding two or three, we call them domains,

but call it like just business areas, processes

where they can really focus to drive the change

versus spreading across the entire organization.

When we looked at it historically, the notion was,

you know, a lot of people were having

what we would call use cases,

which were very small things.

You go try a pilot of one thing,

and what we found was like really interesting,

great insight,

but it didn't scale to anything

and you could never find it on the P&L, right?

So it led to this frustration.

And so the idea of a domain is either an end-to-end process.

Think about something like order to cash

or a procurement organization

or something we can all relate to

as calling a customer service center, contact center, right?

And if you take those domains and you reimagine them,

typically there's 10, 12, 15 different use cases

that kind of bundle together.

But you're tackling something big enough

that it can actually make a difference

and you can find it in the P&L.

I think the second thing we've seen

is really trying to think about

not just making what you do today a little better,

but fundamentally changing a process

or building a new business.

And then the last thing is the people

that treat it like a real business transformation

and not, I'll use the word

just a technology transformation.

You know, we've seen a real difference with people

that think it more holistically like that.

Yeah, that part is so interesting to me,

and I'm wondering if you can unpack it a little bit more.

The idea that it's not just about a company

spending a bunch of money on AI, right?

It's actually about transformation

and catalyzing real change.

Can you talk a little bit more about that?

Because I think that's a really important distinction

that is often missed

when we talk about X amount of money

is being spent on AI,

but it's so much more than that.

It's easy to go spend money

and find things that are interesting.

I mean, we can use in everyday life all the time.

And the same thing is true for all our clients

and all the companies out there.

What we've seen is where people can really go and say,

Hey, here's a business problem

that I'm trying to solve.

And there's a subtle nuance between saying,

Where can I apply AI?

And saying, What is the business problem I need to solve?

And how can AI help solve that business problem?

And when you come at it from that second lens,

you're really thinking about

as a fundamental transformation.

And then it comes with what can the technology do?

What do I need my people to do?

What are the capabilities of the organization?

And it fits together much more holistically

than just I have a new tool

that I want to point in some direction

and not really knowing where to point it.

You also mentioned the idea of focus,

which I think is so interesting in the sense that,

you know, it would be very easy for an executive

to address their company and say, We're all in on AI,

and everyone to then sort of just scramble around,

like wondering what to do about it.

How can leaders identify those areas of focus?

I mean, how do you look at an entire organization and say,

Here, here and here, but not here and here?

I think it's healthy to do the experimentation broadly.

Like people should be using AI

to make their lives better.

I think that's all very good.

But when you think about it

from a company or a CEO perspective,

it's actually not any harder than it is

when you've done any transformation in the past, right?

You got to go and understand what's the opportunity?

How much is it worth?

What's the investment required to get there?

Is it going to be important for me

to transform my business in that regard?

But a lot of people don't actually go through that process

of then understanding what are those two or three areas.

And then making the hard call.

Sometimes it's actually more difficult to say

we're not doing four through six

than it is to say we are doing one through three.

Now, if a CEO took one thing away from Rewired,

from all of this work that you've put into this book,

where would you point them and why?

What's sort of the most important takeaway to you?

You got to be able to ask really good questions, right?

It's amazing.

A CEO or any senior executive can set a great tone

just by asking, did you use the LLM

to check your answer here?

How did you start?

Which tool did you use? Et cetera.

Quite frankly, one of the best things I've seen

is when a CEO takes somebody

that's a more junior person in their organization

and just says, Hey, come show me how you do it.

Show me what you do.

You know, so many times the senior executives

are used to apprenticing the younger

and the earlier tenure folks in their organization.

And honestly, apprenticeship has never been more important

than it is right now.

What is context?

How do you have good business judgment?

How do you communicate?

Those are more important than ever

in this tech, in this AI-based world.

But it's also awesome opportunity

for some of us that have been around the block

a little longer

to have some of our earlier tenure colleagues

share with us because we can learn from them

on a lot of these fronts.

It just sets a wonderful tone

for what the rest of the organization can do.

That's a great suggestion.

I love that.

Now, I want to move over

and ask a little bit about your work at McKinsey.

So your title is,

Global co-leader of the tech and AI practice.

I thought that was interesting

because you're saying tech and AI,

not one or the other.

Why pair them together?

In our experience,

the line between those two things

is either blurring or completely disappearing.

The idea of how do you actually bring

kind of the business enablement through AI

with the core technology

that you need to run your business

has never been more important.

And quite frankly, when we look back

at some of our best transformations with our clients,

you had those two things

just really hand in glove to make them successful.

We talked a little bit about sort of the difference

between spending a lot on AI versus transforming.

But when you look at companies

actually getting return on that investment,

sort of seeing real results,

are there distinguishing factors there that stand out?

We've really found making sure

that the business owners, if it's procurement,

that the chief, the CPO has to really be all in on,

you know, transforming that.

It has to be their initiative, their effort,

not something that's happening to them.

So just that really senior leadership and buy-in.

I think the second thing that we've seen

just super clearly

is a focus on building the capabilities

of the people that are going to be using the tools

and the assets, right?

We've seen this from technology transformations

going way, way back.

There's a lot of good money spent

that sits on a shelf, right?

And so this notion of having the people

that are going to be using the tools

and using the capabilities

fully embedded in doing the work,

making it their solution versus, you know, again,

something that's happening to them.

And then the final thing I would say is like,

it's a very simple thing,

but really tracking, are you getting the benefit?

Are you getting the impact?

And if you're not, why?

Now, I want to talk about Toyota.

I want to get sort of really specific here.

They had this moment, as I understand it, during COVID,

where a VP asked, and I love this question, quote,

If Amazon can track a $10 book,

why can't we track a $40,000 car?

Yeah. Good question.

Walk us through what happened after that.

I love this quote because it didn't start with,

What can the technology do?

It started with, We have a real business problem, right?

And so what they did is they went out

and built a digital twin

that basically replicated

their entire end-to-end supply chain.

So think about the process

to procure parts, make and deliver

two and a half million cars.

And this digital twin was capable of processing

on the order of a billion decisions in 15 minutes.

And what it took was a process

that used to be a hundred people

with, I kid you not, 75 distinct spreadsheets

over the course of a month to get to a plan

and kind of packaged it into this one digital twin.

And what it did was it created

about $800 million of value for them,

mostly through increased throughput,

and then actually getting the car to the right place

at the right time, right?

Because if you want a blue Camry,

having a red Lexus isn't that helpful, right?

And so the idea of actually bringing that to life

was super powerful and impactful for them.

Now, one of the things that stands out to me

from that Toyota example

is that they brought their own people

into that process, right?

Into this idea of building the digital twin.

They didn't bring in consultants

or external technologists.

Talk us through that and why that stands out to you.

One of the things that we've seen

is a failure mode in AI

is when you have someone building a tool

and then hands it to the person

that's supposed to use it.

And especially with AI,

we've all read about AI hallucinations.

We've all read about some of the challenges.

And so it was super important

that everybody that was going to have access

and use this capability had to really trust it.

And so from day one, hour one,

there were sales executives, manufacturing experts,

there were planning experts.

And not only did they help

and absolutely created a better solution,

they were also the evangelists and the change agents

out to the rest of the organization

when it was ready for prime time, so to speak.

So we think that's an absolute best practice

and it was really cool to see in the Toyota example.

Yeah, I think it's so interesting to think about that

in the context of transformation,

which we talked about,

and also of this technology becoming ingrained

in the culture of an institution, right?

Rather than being something

that is sort of foisted upon a group of employees

or a company.

It's something that becomes internal

to the organization.

I love that.

Now I want to turn to talk about agentic AI,

which is obviously a huge topic.

It's something we're covering at Wired,

I would say on a daily basis,

maybe multiple times a day in the last several months,

runs through a ton of your research right now

obviously as well.

Why are you spending so much time on it?

Why is it so different to you

from what we are looking at now

in the context of AI?

So when you look at AI,

I want to say yesterday's AI

because it's still today's AI,

but pre-agentic was a lot of taking

how things work today

and make it faster, make it smarter.

Agentic AI creates an opportunity

to actually change how things work

and have agents interacting with one another.

And so what it does is it takes the notion

of taking a process that's existed

and trying to speed it up

and fundamentally create a new process

where these agents are interacting together

to kind of create a close the loop process.

And so it changes the game on AI,

but it also changes the game

on how people have to interact with the agents

or AI more broadly.

Dan, one of the concepts at McKinsey that you've been

advancing is this idea of humans above the loop

rather than humans in the loop.

Can you explain that to us?

So when you look at more traditional AI,

what would happen is you'd have a human side by side

checking every step of the process

and really actively involved in the do loop, so to speak.

And this new view from an agentic AI perspective

is you actually put the human above the loop.

And so they're more providing context, orchestrating,

checking to make sure things don't go out of whack

versus really being side to side.

So it kind of elevates the role of the human

to be sitting on top of this agentic process.

Now, another concept you've talked about

is this idea of a four rung agentic ladder.

Can you explain to us what that means?

The first rung of the ladder is really about think about

using an LLM to go ask a question.

You get a great answer.

It makes you more productive.

Everyone feels better about it.

The second rung goes more to how you're actually using it

to embed an existing software.

So think about meeting summaries.

Think about auto-fill on a CRM system.

And then the third and fourth rungs get to much more about

either fundamentally transforming a process

or building a new business.

You've also made the observation that most companies

are stuck on rung two.

So why is rung two such a trap?

I think rung two is a trap for a couple reasons.

One is rung two is a place

where it gets completely democratized.

And I think it is only natural

when you see a technology like this

to have an entire company want to embrace it, right?

And things like meeting summaries,

things like some of the ways that you can use.

Everybody looks at it and says,

Oh my gosh, that was super helpful.

I appreciate it.

My life got easier.

I get time to go think about something else

or do something different.

So people fundamentally appreciate it.

It also doesn't take any real hard decision making, right?

You're not telling your team that you're using it

and transforming and my team, sorry,

you didn't make the top three priority, if you will.

So I do think that that notion of it's democratized.

I think we're seeing more and more of a transition

to people, you know, approaching rung three and rung four.

But I think that's the whole notion of Rewired

and how we're talking about this is the real focus

on the two or three domains is a breakthrough there.

So I have to ask you about AI slop.

It's a phrase everybody uses now.

I would love to hear about how companies

who are getting this right are dealing with it.

What works?

For a lot of the clients and companies,

what we see is just being super intentional about it, right?

If you have an agent,

how do you give that agent context?

How do you measure its inputs and outputs very clearly?

How do you track whether it's hallucinating and adjusting?

How do you help it continuously improve?

And if you do that,

what you see is these agents have the capability

to learn over time, right?

And so the amount of slop that's out there

within your organization can drastically be reduced.

I'm also curious, you know,

McKinsey is advising plenty of clients

on the AI transformation, on agentic AI,

but you're also living it internally.

So what have you learned from your own teams?

This is drastically and, you know,

fundamentally transforming the way our teams work.

The currency goes from really good analytics

and for lack of a better way,

like being amazing at using a spreadsheet

to really good asking questions and prompting.

And so how do you actually structure the problem

and get through it with an LLM

or whatever tool you're using

in a way that gets you the outcome you need?

I think it comes back to one of the things

that's been a core part of McKinsey problem solving

and problem solving more broadly for a long time

is really understanding what's the level of conviction

you have to kind of feel like you've got the right answer.

And for us, sometimes that means being 100% right.

Sometimes that means you got to be 80% right

and you got to go test and try and see what happens, right?

And so I think that's where that really good

business judgment still really matters.

Now I had another case study I wanted to ask you about.

So Toyota was all about supply chain.

DBS is a bank, completely different context.

What makes the DBS story stand out to you?

Well, the first thing,

just to compare and contrast the two,

is the domains they picked were totally different, right?

And so it does go back to step one

is understanding what are the priority domains for you.

I think the second thing for DBS that stands out

is how they have thought about it

as a fundamental transformation from day one, hour one,

and gone from round one of two or three

to round two of the next two or three

and so on and so forth.

But they had that vision from the very beginning.

Now you've said that capabilities compound.

So what did that look like at DBS?

What they did is they started

with the two or three priority domains

that they wanted to tackle.

They made progress on that,

but they did it with the vision of

they were then go to four to six

and then seven through nine or whatever it is.

And so the first round took them, call it 12 to 18 months.

The next rounds have taken

more like two, three, four months

because they got smarter at how to do it.

The organization was receptive

and understood how it worked.

And so when you talk about those capabilities compounding,

not only did they start sooner,

but they're building that next set of capabilities

or those next domains that much faster.

It kind of continues to expand

their competitive advantage, if you will.

Now you said something really interesting there,

which is that the organization was receptive.

And that is no small feat.

What did DBS get right organizationally,

sort of on the human side of this,

that other companies might miss?

DBS had a long history of transformation

and technology-based transformation.

And so they had, I think, laid the foundation

in a really positive way of,

Hey, when new technologies are out there,

we're going to adopt them.

It became embedded in how their organization works.

So when some new technology came along,

it wasn't, What are we doing here?

It was, This is the next chapter

of the book we've been writing,

which I think is really powerful.

The second thing is they went right from the beginning

and said, This is a people transformation, right?

Technology is absolutely core to it,

but this is a people and a human transformation.

And so they thought about it,

you know, people back and employee back.

And I think it made a huge difference to their ability

to not just get the first wins on the board,

but then move really fast after that.

So you are a sports fan.

And professional sports is one example.

You know, they often sit right at the edge

of performance analytics.

But I'm curious about what you're seeing

maybe in that world, the world of sports,

and what parallels a CEO can draw from that.

We've found sports to be a wonderful place

to both inspire people,

but also, you know, motivate CEOs.

You know, McKinsey's QuantumBlack,

which is our, you know, leading edge AI part

of our organization, did a lot of its early work

with things like the America's Cup and Sailing,

and has done work more recently

with Team Visma-Lease a Bike, all around, you know,

the cutting edge of how you create that linkage

between analytics, AI, and performance.

And so the notion of creating a AI cockpit

that cuts between a coach and a cyclist

has been transformative.

And you think, what could be so great about this?

Well, you think about the terrain, the weather,

the road, the bike, the tires, all these factors

at the very highest level of performance

drastically influence the decisions they make

minute by minute, second by second.

And so it's really been a wonderful capability,

and it can be a great way for people to understand

how you can use some of these domains

to fundamentally transform in a more creative way.

I love that, I love that.

Now, I know you also spend a lot of time

advising your clients on HR,

so the human resources department in their company.

I'm curious about why that is.

I mean, it's obviously not the department

that people would automatically associate with AI.

Well, I think there's three different parts to this,

all of which have kind of separate but very important.

One is, we talked about it earlier,

this notion of the mindset of the organization

going into a transformation like this.

Getting the communication right

about what you're setting out to do

and why you're doing it is super important.

If people come into it with a negative lens

or a negative attitude,

you can kind of be behind the eight ball

before you even start.

So I think that's a really important part of, you know,

HR or people organization to typically play.

The second thing is we see a huge part of this

being the capabilities of the human beings, right?

How the skills you have to do your job a year from now

are probably pretty different

than the skills you have to do your job a year ago.

And so often training and capability building

are a huge part of what a people

or an HR organization does.

And so I think that's a second important part.

And the third is just the way the organization works,

how things get done is actually fundamentally different

in a lot of situations, right?

And so people that just let that happen naturally

sometimes don't end up in the place that they want to

versus taking it on very intentionally from the front.

This isn't just about us though.

I mean, you're a parent, I'm a parent.

What do you tell people who are worried

about the next generation right now?

The people that are quick to learn,

they adapt when new things hit.

They're not stuck in the way things used to work,

consistently do better, right?

Being a creative problem solver,

being able to communicate,

being empathetic and relating to other human beings.

All of those things are things that we think will be true

no matter what the technology continues to do.

My advice, and by the way,

my kids would laugh at me giving advice to other parents,

I think, is how do you find that balance?

We want to immerse them in the technology,

but we also want to have them learn the things

that we all learned, you know?

And some of that is creativity, self-play,

some of those things that you just don't learn

in front of a screen.

And so what I talk to my kids about

and what I talk to my clients about are,

can we be resilient and can we be adaptable?

And if we can be those two things,

I think we're going to be set up for success.

And what about the CEO?

If a CEO takes one thing away from this conversation,

what do you hope that it is?

Well, I'm going to cheat

and I want them to take away two things.

One is the technology is pretty remarkable

and you need to embrace it.

The second is your role has never been more important.

And the ability to think about this

as a business transformation

and really set that expectation

with your entire organization.

Now, five years from now,

when we're looking back at this moment,

what do you think we will all be able to see very clearly

that's maybe hard to see right now?

I think we'll be able to see

that the change is actually durable, right?

Is the technology going to last

and is it going to fundamentally change what we do

and how we do it?

I think the answer is affirmatively yes.

And therefore the implication is I think companies

that embrace that now

and really think about what are the lessons

and how they get out in front of that,

I think that will also be a durable advantage

that we'll look back on five years from now.

Now, Dan, if you will indulge me,

we always close these conversations with a little game

that we invented at Wired.

You're like, Yeah, they warned me about this.

It's called Control, Alt, Delete.

And I want to know what piece of technology

would you love to control?

What piece would you love to alt?

So alter or change.

And what would you delete?

What would you vanquish from the earth

if given the opportunity?

The control, I will say, is actually, you know,

self-driving capabilities.

I spend a lot of time on the road to clients,

driving kids to sports.

And the ability to make that productive time

would be a game changer for me.

So I'll take that as one.

The alt will actually be one of the things

we were talking about.

I would think about altering kids' apps,

and that's a broader than just one.

But I think getting this balance right

of being really savvy in AI technology

and having kids that, you know, go outside

and don't have any structure

and play for three or four hours

and learn how to be creative,

learn how to work with peers,

learn how to interact with others.

I'm cheating a little bit on my answer,

but kind of alter how the kids interact with screens.

And then the delete, I'm going to go back to my sports.

I was a tennis player in college.

I'm still involved

with the International Tennis Hall of Fame.

So like the legacy of history and everything

is super important to me.

And I love the automatic line calling for tennis

when it came out

because you're never going to get a line wrong.

It makes me kind of sad not to have, you know,

line umpires, men and women on the side

of professional tournaments anymore.

I didn't know that the men and women

had been eliminated from the tournament.

Yeah, so now it's just all automated.

So there's that makes me a tiny bit sad.

Because they can't have the tantrums

and like throw the racket around anybody.

Exactly, I mean, we all grew up loving John McEnroe.

Totally. Yelling, screaming

and brought it to life.

So we'll miss that from now.

It's harder to argue with the screen.

So that'll be the one I'll delete.

That's a really...

I've never heard that one before.

I will give you that.

Dan, thank you so much.

This was fascinating.

Really enjoyed the conversation.

Thank you so much for having me.

Really enjoyed it. Excellent.