July 15, 2026 • 51 mins
"AI doesn't have judgment in it — it produces judgment-shaped artifacts."
Usman Sheikh, founder of High Output Ventures and author of the Framebreak newsletter, has spent years writing about how AI will reshape professional services — and the world has now caught up with the puck. In this conversation with Aidan McCullen, he explains why the consulting pyramid — the structure behind every major firm — is coming apart, and what replaces it.
In this conversation, Usman reveals:
- Why firms became pyramids: leverage, margin, and a talent academy paid for by the client's billable hours
- How AI pulls all three apart at once
- What "judgment-shaped artifacts" are — and why AI is fundamentally different from offshoring and RPA
- "Fluent but substitutable": why leaning on an LLM turns your career into a slot machine
- The digital twin trap — why a frozen snapshot of a senior partner decays fast
- The Concentrix vs Sierra story: AI voice agents in regulated industries growing from $100M to $200M in revenue in four months
- The "AI discount": why KPMG asked its own auditor for a price cut — and how the surplus drains to clients and platforms
- Why Palantir and Microsoft are suddenly preaching sovereignty over your data and models
- The intelligence core: the feedback loop that turns pyramids into rings
- How firms build tomorrow's bench — simulators, shadowing, apprenticeships — and why partners near retirement have no incentive to bother
Chapters:
00:00 Sponsor Message
00:31 Guest Introduction
03:10 Why Firms Became Pyramids
05:13 AI Breaks The Model
05:53 Digital Twins And Hiring
07:38 Training Juniors With AI
11:50 Fluent But Substitutable
15:19 Offshoring To AI Shift
20:18 Macro Jobs And Demand
23:14 Who Captures The Gains
28:04 Learning In The AI Era
34:33 Intelligence Core And Rings
40:49 AI Discount And Value Drain
45:21 Building The Future Bench
50:11 Wrap Up And Where To Find
51:16 Final Thanks And Outro
About Usman Sheikh Usman Sheikh is the founder and managing director of High Output Ventures, a venture studio building and backing technology-augmented service firms, and the author of Framebreak, a weekly newsletter on strategy, consulting and the knowledge-work economy.
Read his work at https://framebreak.com and connect on LinkedIn: https://www.linkedin.com/in/usmans
About the Host Aidan McCullen is the 2025 Thinkers50 Innovation Award recipient, a keynote speaker on AI, disruption, innov
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Before we begin, I wanna thankour sponsor, Kyndryl, who run
(00:02):
and reimagine the technology thatdrives the world's infrastructure.
We're talking about healthcare, we'retalking about aerospace, and banking
amidst many, many other i- industries.
But also check out the KyndrylInstitute where you'll find
unbelievable thought leadership, manyof the guests we've had on the show.
You can find both at Kyndryl.com,Kyndryl.com/institute or .com where
(00:28):
you can find out more about Kyndryl
today's guest is a previous guest froma webinar we had recently with the
Kyndryl crew, and it's great to have him.
He's been writing lucidly about AIfor several years, and particularly
how it's gonna affect knowledge workand professional services companies.
was saying to him before we came onair, it was a little bit like that quote
(00:50):
where Wayne Gretzky was asked how doeshe always know where the puck is gonna
be, and he said he skates to wherethe puck is going, not where it is.
And that's a little bit what'shappened with our guest, because
he's been talking about this stuff.
Earlier on, people thought maybe,"Is this gonna happen? Is it real?"
Now they're listening, 'causethey've caught up with the puck.
He is the founder of High Output Ventures.
(01:12):
He's the author of a brilliant sitecalled Framebreak, Usman Sheikh.
Welcome back to the show
It's great to be on.
Looking forward to this conversation.
Always enjoy these where, like yousaid, the world of professional work is
changing, and there were signs last year.
But lots of people have talked about thedeath of consulting for a very long time.
(01:35):
and by no means do I think, like, there'sgoing to be the death of professional
work, but I definitely do think thatthe way that the firms were structured,
the way the work was delivered, isfundamentally going to change, and that's
what I'm most interested in writingabout and talking about these days
I was having breakfast recently with asenior leader, and I was saying to him
(01:56):
that there's two concurrent shifts ormultiple shifts going on, but the, with
regards to firm, was saying to him, I, wewere having breakfast, so I picked up a
salt cellar, and I was like, " the firmused to be like this, like an IBM." And
then I pointed to the gr- grains or theflakes of salt that were on his avocado.
And I was like, "See the flakes ofsalt there on you?" I said, "So the
(02:18):
firm, like IBM, had to break up tobe more agile and became modular.
then eventually by doing that, it invitedin lots of competitors, and then I poured
out some salt on the table, and he waskind of looking at me going, "What is
this guy gonna talk to me about?" AndI was like, "Now it's like grains." I
said, "But AI is ta- " I took some ofthe grains and I put them in my glass
(02:41):
of water, and it's like, it's dissolvingsome of those grains, and it's leaving
some, it's leaving only very few.
in there he went, he was looking atme and was like, " did that work?
Did you understand what I was sayingthere?" And he goes, "That's a
brilliant way of understanding it."And I was like, "Oh, yes." So I was
like, you've written then about thatstructure and new structures that
(03:02):
are needed, and I thought maybe wewould start with why was the structure
the structure in the first place?
And then we'll get ontowhere you see it going.
I think professional service firmssolved a-- the way that they were
structured solved a scarcity problem.
The firm's most valuable assetwas its senior leaders' judgments,
the senior people at the top.
(03:24):
But the number of hours in theday were limited and, you know,
they could only service a certainnumber of client requests.
In order to solve this problem, whatthey did was that they put partners
at the top and built layers ofmanagers and juniors beneath them so
that they could scale their judgmentacross a much wider array of clients
than any single one person could.
And I think that became theleverage unit that we see today, and
(03:49):
that's why the structure existed.
And more than just multiplying the outputof the senior leader, what it also did
was that the base of the firm was theprofit engine of these firms as well.
So the rate that the firm wouldcharge for the work versus the
rate which they paid the base, thedelta between that was the margin.
(04:12):
And the wider the base became,the more money the firm made.
And also the, the people at the base ofthe firm were actually very profitable
units of margin for the company, andtherefore it expanded its way outwards.
So two things came outof the pyramid so far.
One was leverage of the seniortalent's judgment, and the second
(04:34):
was the profit and the margin.
And the third thing, which I thinkis not talked about as much, is
that the pyramid also was a trainingground for the next generation of
leaders and signers at the company.
So when you have a wide base, you canfigure out who is good, who is doing the
work well, and through a series of errorcorrection, supervision through seniors,
(04:58):
these junior people ultimately becamethe people who could one day put their
name on a document and sign it as well.
And the beauty of this was that all ofthis training was paid by-- for the client
under the guise of the billable hour.
And what AI does is that it pulls allthree of these factors, the leverage,
(05:22):
the margin, and the formation of talentaltogether, which then puts a lot of
pressure back onto the firm becauseif they are to adapt, they're going
to have to figure out how to changea model that has worked remarkably
well for them for a very long time.
I love the idea of the pyramidand the base and the understanding
(05:44):
there that a training ground.
And then it opens up the question,and we'll… I- I go all, all
over the place, man, here.
You know
Right.
roll.
But I also fear that, we've seen thiswith all these firms, that they're
not hiring for that base anymore.
So I was thinking like, "Well, whatare they gonna do then?" obviously
(06:05):
training AI you know, there's somefirms bonusing people on using AI
or punishing them for not using AI.
But then on top of that, I've seenthis concept where if you have a senior
executive or a senior consultant fromone of those firms, they're creating
a digital twin version of that person.
Right.
And
Yeah
(06:25):
may be let go in the next few months oryears, but there's this digital twin.
So you have frozen that knowledge at thestart, at the top, and then maybe actually
filling in through outsor- sources ofinformation, new papers, et cetera, your
future work that you're about to publishin the future, stuff like that feeds it,
(06:46):
and then you have this, just a- a- a- a-an LLM perhaps that is owned by the firm.
Yeah
know that's n- not exactly theright language to put on this,
but that's the concept that we're
Yeah.
There as well.
Yeah.
So I think that youcorrectly outlined two parts.
One was the digital twins of seniorleaders, which we can park on one side.
(07:11):
And the other part, which I think appliesmore to the formation base, which is
that if you stopped hiring as manyjuniors into the firm, you don't create
a problem today in the market becauseyou can hire from other firms and other
people from the stock that exists today.
But maybe five years from now orseven years from now, the market
(07:32):
is going to face a glut of thattalent which does not exist.
So I think I really thought about the,the formation problem quite a bit, and
let's talk about the junior problem first.
So the first thing that I said was thatthe base of the the formation was paid
for by the client in the traditional modelbecause, you know, you learnt on the job.
(07:57):
Now, when the client, let's say, doesnot get billed for hours of work,
and they just get billed for, "Thisis the outcome that I would like,"
and you have to figure out how tobuild that for them, you take that
abstraction away from the junior base.
And now what the firm has to do ismake an intentional decision as to
(08:18):
whether they are going to invest intothe formation of talent in the future.
And I'll tell you from personalexperience, we have a IT services
company in the, the portfolio, andthis is a discussion we regularly
have because it just takes a longtime to grow that person, nurture
(08:40):
that person into that pers-- into thatperson who can sign in the future.
And the decision now is (08:44):
Shouldn't we just hire someone from the
market who is mid-tier and sort oflike pay them a little bit more,
but augment their output with AI?
And I think this is a veryintentional decision that someone
is gonna have to, to make.
Because you know, I have friends whoare working on building simulators
(09:06):
for junior talent at some of thesefirms that, okay, so if you're not
doing the work, there are simulators.
In one of our firms, what they make thejuniors do is that when the output comes
out from the LLM, the junior has to assessthe output, and then without seeing what
the senior or the system would recommend,and then they cross-triangulate,
(09:28):
and they're doing all of this work.
So all of this stuff is happeningright now, but who's paying
for all of that development?
It's the firm is payingfor that development.
And I think, like- It's the correctchoice for the firm to actually set
some budget to do this because theyset themselves up for durability of
(09:48):
competitive advantage in the future.
Because let's say that we do allof this and we skip the thing, five
years from now, we don't have a newstock of signers, that we can't afford
new signers or senior people in themarket anymore the firm unfortunately
may have harvested profits for fiveyears, but is going to find itself in a
(10:08):
really tough spot five years from now.
So, so I think that's a formation thing.
The core, if it is that intentionalityor talent formation becomes the
responsibility of the firm and the payoffis much further out, but it's perhaps the
correct decision to make in the short term
(10:29):
I see it exactly the same as a sportsteam and exactly the same as making
some bets in innovation that m- manydon't work out well, but a few may pay
for all the ones that didn't work out.
But back to the sports idea, r-because we're in the middle of the
World Cup, I'm gonna age our episodenow, but of course, it's about AI,
(10:50):
so it's gonna age very quickly.
But I recently said this aboutRonaldo, so Ronaldo, brilliant player,
absolute unbelievable respect to himfor playing so late in his career and
keep going and be able to perform.
But there's lo- huge questions overshould he have been playing, and for
me, what that means is he's preventingsomebody else from developing, and as
(11:12):
a result, and this is the choice ofultimately the coach or the management
team, preventing others from growing.
And I often think about that.
If you have a digital twin, say,of a senior consultant, and you're
relying on that versus to what yousaid, where I have somebody else going
through ultimately the struggle that'sneeded to actually learn, 'cause it's
(11:34):
through the struggle of learning.
We were talking about some of thepapers we read on AI before we came on.
I mean, it's a huge struggle to readthose papers, to get through, and actually
new terminology and terminology you'reunfamiliar with, but it's only through
the struggle that you actually learn.
Right
one of the things I wanted to say wasthis concept of the Chinese room that
(11:54):
we talk about in AI, where you actually,like, push a piece of paper under
the, the, f- under the, beneath the,
Yeah, a dork
and the AI does the work andpushes it back, and you've no idea
what happened be- after that whenit was pushed through the door.
That more and more, even if youare a junior coming into the bottom
(12:15):
of that base layer of the pyramid,many are using AI and not going
through the, the, the disciplineand the struggle of the learning.
And in there, you're like, don't know whoactually knows what anymore because all
you're ultimately selling is the output.
And it, it's, I, I think the governance ofthat is huge, and it can be missed so much
(12:37):
because you think you're doing the work,but it's actually the machine doing the
work all the time, and nobody's learning.
Yeah.
I wrote a post about this, and Icalled it fluent and substitutable,
which goes to the exact word that youjust did that you use the machine.
So I used an example of a firm where the,the junior was struggling to r-get to a
(13:00):
certain outcome, and they were relyingheavily on the machine to produce outputs.
And they thought that it's justsome iterations that I need to
do, a few more button presses,and I'll get close to that number.
And it doesn't work that way becauseyou've just converted the LLM into
a casino or a jackpot machine whereyou're just hoping that the next button
(13:22):
press is going to give you the answer.
And while it could happen, Idon't think anyone bases their
paying their rent on whether theslot is going to pay out or not.
And similarly my advice to that junior wasthat you have to do the work by hand to
understand what the machine is doing sothat you can actually grade the output.
(13:44):
And if you do those difficult reps, youactually understand what's happening
because if you don't, then everyone whocan use your same prompt or eventually
the machine will just absorb yourwork because if you don't know what's
happening, the machine is like, "We'reboth guessing over here," and the
machine gets more reps at it than youdo, the machine's gonna win at the end.
(14:08):
I was thinking of that movieWALL-E, you know, the, the I
think it was a Pixar movie, and
Yeah
for watching Pixar, but one of theanimation studios, WALL-E, this,
about this robot, and it's a worldwhere humans don't do very much and
we're all extremely obese 'cause we'veoutsourced everything to the machines.
And I was like, oh, well, that'skinda what a lot of people are doing.
(14:29):
Where I was talking to a, a teamrecently, and they were saying that if,
for example, throw this out there, thatyou're creating policies, huge policies
for huge organizations, but actuallyyou're almost just the person punching
in the data at the start, and it's themachine coming out with the policy.
(14:50):
So nobody actually understands the policy,and everybody's working towards those
policies, and you're kinda going, there'snuances in there that are gonna be missed.
There's gonna be somany thi- things missed.
There's also gonna belots of stuff spotted.
But in there, there's a huge threat.
So maybe so- some thoughts on thathappening, 'cause it, it is happening
all over the world, before we gointo talking about business process
(15:12):
optimization, outsourcing, offshoring
Yeah
of the professional firms use, andwhere that fits into all of this.
I think when we look at past moves suchas, like you mentioned, offshoring.
Offshoring essentially changedwhere the work got done.
It didn't change anything else.
It basically changed the jurisdiction,lower cost jurisdiction, and the
(15:36):
work was still done by people,but it was done somewhere else.
Workflow software, basically RPA and otheraspects, made predictable processes into
a structured format within encoded rules.
So it changed how you did the work,but within a certain structure.
(15:56):
And I think when I think about AI,what it does is that it takes this
reusable system which now can freezeitself from this constraint of rules
that were around and it changeswho is doing the work now, right?
So it doesn't produce… WhatI've talked about is it produced
judgment-shaped artifacts.
(16:17):
Like, you know, it's-- it doesn't havejudgment in it, but it's ju- shaped.
Like someone who doesn't know what agood brief looks like will probably
think AI has created a great brief,but someone who's created briefs
for a living will know, like, "Okay,this is not a bad brief, but it
made mistakes in these four areas."
And I think, like, that's how I seeAI being very different from the past
(16:40):
world, because in the old models, itwas still when the machine reached
or when the process reached a certainexclusion that it couldn't understand,
you always had to revert back to a senior.
But now that has expanded, and you needfar fewer people sort of like doing that,
and the machine starts everyone off notfrom zero, but from later in the cycle.
(17:03):
And I think that's fundamentally differentfrom past shifts because it has direct
implications on who's doing the workand how the work gets done compared.
Because past shifts, a lot oflegacy co-owners will say fortified
the pyramid more than anything.
They became bigger, like, you know,Accenture has like eight hundred
(17:26):
thousand people, a lot of them in,um, you know jurisdictions like
India and Philippines, et cetera.
But AI does not necessarily play outthat way because if you just use it as a
productivity enhancer, um, I think yourlong-term competitive advantage deflates
(17:49):
very quickly because everyone will usethe same tools to get to the same answer
You remind me of an old story ClaytonChristensen told in, I think it
was How Will You Measure Your Life.
He, he talked about the, he called it theship of Theseus, but the whole idea of the
ship of Theseus, for those who don't know,is, like, the ship gets replaced plank by
plank, and eventually you're kinda going,"There's no planks from the original ship
(18:13):
so wh- what, is it still the same ship?"he talked about, well, Dell outsourced
more and more of one of their machinesout to China, I think it was, with ASUS.
And ASUS then eventually, as Dellkinda went down the stairs from
their side, ASUS were going up thestaircase from their side, building
ta- capability more and more and more,and then actually becomes a competitor.
(18:37):
And we've seen thatwith BPOs, for example.
And actually, where you probably see itmore, and I haven't seen anybody really
call this out, is for example, of theEuropean or even a lot of the US companies
maybe use India, and now the knowledgework in India is absolutely amazing.
(19:00):
The people coming from India are amazing.
A lot of the new mortgages herein Ireland are actually Indian
people who have moved to Ireland.
So
Yeah
Kinda reversals of that, the sametype of thing of the outsourcing.
Yeah
in there is a huge threat.
So we talked about the threatof on-boarders, new starters
in some of these firms.
(19:21):
But then think about the other side.
If y- if you've a BPO out in India orwhat used to be an emerging market,
and I don't need that anymore,
Yeah
ecosystem, a huge house of cardsthat could come tumbling down.
Because somebody told me, you know,if you're working for one of these big
(19:42):
firms in a BPO over in India, you'reextremely well paid by local standards.
You're getting Ubers to work.
You're living in luxury locally.
But if that's gone now, not onlydo you fall, but so does the actual
house of cards that's dependent on you
Yeah.
And it's a major source of employmentwithin the, the country's base as well.
(20:06):
And a lot of their exports are directlylinked to this export-based revenue that
gets collected back into the country.
So it has large macro implicationsfor the country as well.
And there's this large narrativeshift of like, okay, well, we have
all this trained labor, like you said.
(20:27):
We're going to convert them intothese forward-deployed engineers
and, you know, suddenly they're goingto do this other new types of work.
And the reskilling narrative isdefinitely there, and there's also the
conversation that as the professionalservices work will get cheaper to produce,
(20:48):
then there will be a lot more work andtherefore, hopefully, a lot of these
people will find employment within that.
But I think there's two things in thatargument which I wanna click into.
One is like, invariably, if the answersget cheaper, you're gonna have more
people asking more answers for it.
(21:09):
So, so that's there.
But there are two things moving over here.
One is the demand for that unit ofwork, and the second is the human
labor cost to deliver that unitof work, however much is required.
And so if the demand rises faster than thehuman labor cost rises, then employment
(21:32):
will, will rise because you still needmore humans in order to fill the thing.
But if the human cost falls fasterthan the demand sort of like rises,
well, then the reverse happens and, youknow… And I don't think it's going to
be clear-cut across the entire industry.
I think serv-- certain service linesare going to still be heavy human labor
(21:55):
required, like high-stakes work is notgonna suddenly be handed over to the
machines because of a lot of things.
But low-stakes work, which is heavilycheckable, yes, I think a lot of
that work is going to be a threat.
And the call center industry is a verygood sort of like analogy of this.
(22:15):
There's a firm called Concentrix,and then I usually compare it to
a new firm called Sierra, whichprovides, again, voice agents.
One uses human labor to pick up the phone,and there's a certain cost towards it.
And Sierra is multiples inmarket cap value, and its
revenue is growing exponentially.
(22:36):
Like, I think they did a hundredmillion they reached a hundred
million in February, and they-Crossed two hundred million in June.
So it's basically the curve isexponential, and what they provide is call
center agents for regulated industries.
And there are no humans involvedin this transaction anymore.
(22:56):
And I think, like, that's where evenwhen the stakes are higher, you can
see the models moving into this.
This is just not picking up the phoneand booking a reservation for you.
This is someone calling up a bank andasking them a complex question, and that's
still being handled by an agent today.
Just goes to show you that it's notnecessary that increase in work is
(23:18):
necessarily going to lead an increasein employment because value capture
can either go to the person, to theplatform, to the client, or to the firm.
So… And that's the hard part inthis conversation that I'm finding.
Like, who captures all of thisgain that is going to occur now?
I, I really wanna talk about that,
(23:39):
'But just on that, what I thought about that before is it's almost
like, I call it this idea of musicalchairs, that game we used to play
as kids, where the chairs run out.
And you think about each shift we've gonethrough from digital to cloud to today,
there's been shifts, but I just don'tsee where there's enough chairs at the
(24:03):
table for people to be all employed whenAI is taking so much of the value and
Bye.
optimizing so much.
And as, as, as to, as what we talkedabout before, it's taken away that
bottom layer, so there's actually
Right
at the bottom layer g-starting, so there's no way
to get onto the layer at all.
Yeah
point there, the valuation of the firmis based on usually revenue per employee.
(24:27):
So one of the examples is Base44,this company that sold to Wix for
80 million had eight employees.
One shareholder, eight employees.
And you go, well, w- if that's thecase, and that's happening more and
more and more, and we're gonna see,as I said s- to you before, not two
pizza teams, but two slice companies,where there's enough just to feed
(24:50):
one person who owns those companies.
And it's a lot of the companiesyou've invested in, co-created.
Right
does the employment lie?
And, you know, people are kinda going,"Oh, well, I'm gonna get my kid to
become a carpenter or go into thetrades." You're kinda going, "Yeah,
but who's gonna buy the houses andemploy them?" So th- there's a whole
ecosystem here needs thinking about.
(25:13):
And the irony, I thought aboutthis, the irony is somebody in
government somewhere where he's kindagoing, "This is on my desk now,"
and they're typing it into an LLM.
Yeah.
Yeah, I think yesterday it, it was GoogleDeepMind founder Demis who wrote an
article on X in fact talking about howthis is this massive change happening
(25:37):
and how we need a collective responsein how to deal with the repercussions
that this is going to have on society.
Because I think the, the ju-juryis still out on is there gonna
be net employment increase?
How are we gonna distribute this?
I've heard arguments on both sidesof the, the thing, and usually
(25:59):
you're arguing your priors.
So if you want more employment, thenyou're gonna say, "Of course, more work
is gonna create more people," becauseif my margin unit is dependent on that
unit, then I do not want less becauseit creates a lot more problems for me.
But if I'm a technology-augmentedfirm which is using this, I'm
going to say the, the, the latter.
(26:22):
Ramp, in fact which is this card--expense card plus now accounting software
company, wrote a paper recently whereit found that high growth companies when
they adopted AI aggressively actuallysaw headcount rise within the firm.
But I think, like, what the papersays in one line was that this is
(26:43):
limited to high growth firms in sortof information technology sectors.
And so your, your concern is valid, andI think a lot of people are seeing that.
It's just very difficult because Ithink jobs are a bundle of tasks, and
what we spoke about earlier was thatif you unbundle the job, there are
certain parts of the job that sort ofget automated away, and the parts that
(27:09):
remain require you to either have highjudgment or the ability to have some
credential to stand behind the answer.
You might have a, a, a law background ora audit certificate or a CFA or a doctor.
I think that provides you position,but someone who doesn't have those
(27:29):
gets put in a very difficult spot.
Um, and like the first work that we weredoing, sort of now firms need to invest
in you to build those skills where it'snot a byproduct of how I do the work.
That in itself creates, um, tensionwithin the firm because do you optimize
for the short, mid, long, or altogether,or how do we do, or it's very uncertain.
(27:54):
I don't want to invest now.
Maybe I'll invest two years from now.
And these are all conversationsI'm having with portfolio
companies, and it's, it's not easy
You know, I saw this shift when I, Iwas a professional rugby player, as
you know, and I went into the workingworld in 2008, which was terrible
timing if you look at it that way.
(28:15):
But I was, I actually purposely went inat that stage because I was like, "Oh,
well, this is gonna be difficult, andI may as well go into an environment
that's difficult for everyone," andused the kind of perseverance and the,
the resilience that I'd built up fromsport to work in the working world.
But one of the places I first saw wasI saw salespeople really struggling be-
(28:38):
because before they'd actually workedas booking agents, or they were esser-
essentially just minding their clients,and now they have to actually sell, and
they'd actually forgotten how to sell.
And before we, we talk, 'cause Ireally wanna share the shape of the new
firms, you talk about five differentstructures, shapes that a, a firm can
have and the evolution of those shapes.
(28:59):
But the, the atrophy of those skillsif you don't use them, and one of
the things I see people, and thisis why I was talking about that
mental atrophy, that you may becomementally obese because you're not u-
actually using that muscle anymore.
There's a huge onus on people to becomeconstant learners or be lon- constant
learners, and I know it's one of themotivations behind Framebreaker through
(29:23):
your writing, my writing as well.
It's not only for the audience to sharethis stuff, but it's actually for me to
actually understand it, and going throughthe struggle of figuring out arguments,
does that work, does it land, my
Right.
my salt, metaphor at the start.
D- does it make sense?
And actually going through the processof that, thinking, sleeping on it,
(29:43):
thinking about it over several days.
And when people are using agents moreand more and LLMs more and more and
co-working more and more, the- allthey are are this person who shares the
information that they've just got fromthe LLM, even to a point of, "Upload
the PDF, give me the main points." like,when you read a book I, I used to, I
(30:04):
used to get this from people to go,"Oh, why don't you just use one of these
summary book- you know, like Blinkistor one of these tools to summarize books
for you instead of going through it?"
And I kind of go, Whenever I did that andthen I read the book, I used to do both,
I'd read it and then read the book, I'dpick out totally different highlights."
Yep
much of the book was so personalto m- my own experience, my prior
(30:27):
information that I'd sourced, that I'dbuild information totally different.
So if we all use the same agentsto produce the same results, to get
the same highlights, no innovationin there 'cause there's no spotting
anomalies and information that'sappealing to you or something that
might spark a metaphor for you, and Ijust think that that's a huge shame.
(30:50):
And I wondered what your advicewas for people who are… 'Cause
we're all doing it a little bit.
Yeah.
towards that.
What is your advice?
What, what are you seeing as a bestuse of the tools that are out there?
Yeah.
I like the, the way that you saidthat, you know, you were doing both.
I sometimes use that as afilter of whether I want to
(31:10):
read this completely or not.
Because also, to be honest, a lot of booksthese days, especially the newer books
these days, they fill a lot of pages, butthere are a few ideas that are stacked.
So you wanna figure out whetheryou wanna go through them or not.
But I completely agree with you thatthere are certain books where just the
(31:31):
highlights take all-- they abstracteverything away from the, the fabric
of how everything stitches together.
That if I just-- If you saw Alexander'sBusiness Model Canvas, and you
just saw that snapshot and yousay, "Okay, here are these boxes."
You put your firm on this, and then youtry to figure out w-- how to maneuver it.
I think, like, you know, it's, it's quiteshallow, but if you wanna understand
(31:55):
what is the, the, the core behindeach one of this, how do they each
interact, what are the different parts?
I don't think that you get that.
And that's how I, that's how Itend to read these days as well.
I think certain older books thatyou'll just read through because
there's just so much in those books.
But especially when I read newer books,I do this where I go to the parts which
(32:16):
is also most relevant to me so that I canpick that up, and then if that grabs me,
then I'm reading a lot more of the book.
And but, you know, I, I don't knowwhether the, the audience will see this,
but this Mike Kirsten had written thisbook called Project to Product, and
now his book just came out yesterdaycalled Output to Outcome, which is very
(32:38):
relevant to what we're talking aboutover here because we're moving from a
society which is was driven a lot byoutput, like, especially software firms.
Output was the code and everything, butnow we're talking about these outcomes.
But how do we reorient thefirm around the outcomes?
And he has a structure in thebook, and I read a synopsis of it.
Someone that I really respect r-recommended the book, and here is the
(33:01):
book, and I started reading it thismorning, and it was a great sort of, like
read so far because it connects with a lotof things which are relevant to me today.
So that's, that's my processas to how I'm doing things.
And it's becoming harder becausethere's just so much to read
these days that you have to figureout what's really important.
(33:21):
Seth Godin years ago saying that he, he'dput the summary… You know, the way you
see at the end of a chapter, it usuallygoes chapter takeaways, that he'd put
it at the start of the, of the chapter.
Yeah
through learning w- how the bestoptimization of learning is actually
you, if you have a, a concept of whatyou're gonna learn early, then you
(33:43):
kind of read to almost validate that.
But if you're led down the wrongpath, so this is what I was saying
about the AI could actually lead youdown the wrong path and go, "These
are the salient bits. This is what'srelevant," and then you read it.
You're missing therichness of it, you know?
And, and when I write, I, want to bringthe jour- the reader on a journey.
So actually, in a way,you're warming up their mind.
(34:06):
Like, the way I see it is the metaphoris creating, like, Velcro, and then
the idea, which sometimes is complex,is what you throw against the Velcro,
and it, it sticks to the metaphor, and
Yeah.
Yeah
d- you can't shortcut that.
Like, that has to go
Yeah.
that journey, and it's like, when I runa workshop, I, I kinda do the same thing.
(34:27):
You kinda gotta go through and bring themon a journey, not too long, but bring
them on a journey where, where it sticks.
But let's, let's talk about the, themove of what y- the way you s- you
phrased this, and maybe we'll qualifyand share some of what you mean by this.
So as leverage moves towardsan intelligence core,
(34:47):
Right
then different structures ofthe firm emerge, and there's
five that you identify.
So maybe let's talk about what, what doyou mean by leverage, what do you mean
by an intelligence core, and then these
Yeah
shapes that emerge
Right.
Let me go through it quickly.
We talked about leverage firstwhen we talked about the pyramid
and traditional structures, whichthe source of leverage was labor.
(35:10):
If you wanted to do more work, youwould add more labor, and that is how
this firm progressed or grew bigger.
Now, the intelligence coreis essentially taking the
learnings of how work gets done.
So let's make this real.
A client comes, and they want us toproduce ads for their e-commerce brand,
(35:34):
and we have to put them up on Facebookfor them and test it out for them.
So let's say that we tell theLLM to generate us, "Here's a
brand, here's all this context.
Generate some ads for us." So theLLM will push the ads out, and now
someone with knowledge of both thebrands and advertising will say like,
(35:55):
"Okay, let's edit this, let's trythis, and let's put this to work."
They'll put the ad. There'll be afeedback loop where the ad platform
will say, "Oh, this ad performed well,didn't perform well." Let's say that the
ad came back, it didn't perform well.
We made three tweaks to it,and then the ad did better.
Okay, so that's a simple loop thatI just created over there that, you
(36:18):
know, you could replace that with taxor anything else, or a legal contract.
Most of the times, you just log thewinning ad variation and say, "This
ad did better for every, And it'sjust logged in a library or a best
practice or a template or whatever.
An intelligence score is when the outputwas created by the machine, someone
(36:41):
with expertise was able to make errorcorrections to it, and then was able
to assess what happened with it or not.
And then they would feed all thoseerror corrections back into a
machine which learns all of this.
So, so for all the things that youwould discard, this is putting all
of that context back into the core.
(37:02):
And now the next time the machinesays, "Oh, here's a health brand which
is targeting the same demographic,"the next iteration that the machine
will produce will be of higher qualitybecause it has some learning feedback
loops from before that makes it better.
This is when you don't rely on aperson for a source of leverage, you
(37:23):
rely on the machine to produce thisartifact which abstracted that thing.
This also goes very close to your digitaltwin example that you used earlier
because, you know-- But there's a keydifferentiation I wanna make here.
Digital twin is a snapshot ofyour state of mind today, and that
(37:46):
decays at a really fast pace inthe world that we live in today.
The loop that I'm talking about isthis constant flow of information and
work that is happening at the firm,error corrections that are happening.
And let's say that we've never workedwith a shoe brand targeting women in
Europe, and today we're gonna do that,but we haven't created ads for it.
(38:08):
So someone is going to set that up first,do that, put it into the thing, and
the core becomes smarter and smarter.
Once this works really well, and you canstart to orient the entire firm using
this to ask it, "Here is the client. Weproduce this outcome. Can you produce this
(38:29):
for me?" And it produces output for them.
This loop creates, the feedbackloop happens, and the firm changes
structure because instead of addingan additional person to the team, you
can scale work depending on how muchdemand you can generate and how quickly
you can verify the work coming out.
And the firm could potentially morph froma pyramid to what I'm calling a ring.
(38:53):
And the ring would have theintelligence core at the center,
and then you would have this newstructure of people around it.
And in a nutshell, that'sthis transformation.
And I go through in, in detail forhow it could happen to various models.
But I think as an abstraction, thisshould provide you with an analogy of
what I mean by the leverage, intelligencecore, and how this new shape comes about.
(39:19):
It's funny you were saying thereabout the, have a snapshot of
someone's, essentially their mentalquiver at a certain point in time.
But what you see also is, you see peopleget to a certain level in organizations
and actually stop learning, and actuallythinking, "I, I have it now. I'm here
now. I, I, I've, I'm in port." Andyou're kind of going, well, there
(39:40):
is no safe port or safe haven with
Yeah.
anymore.
You have to consistently learn and finda way to learn, which is why learning
how to learn is, is so important.
And unlearn stuff as well because I thinkthat's where, where people have dug their
own grave in a way with AI, in that y-you gotta be the one who's constantly
(40:02):
provoking yourself to learn, and it,it's uncomfortable, and it's a pain.
Like, I would say, like, some ofthe papers we read are boring.
Like, they are tough academic reads.
They're tough to get through.
But it's only by doing that thatI actually can learn and, and
challenge your own thinking as well.
But you, you mentioned there,so the, the shape changes.
(40:22):
The, the gain then shifts.
And the, so you, you were saying earlieron about, okay, the, the, amount of w-
effort I put in versus what's output is,is, is my profit, essentially, my margin.
But there's a surplus created byAI, and then maybe the client's
(40:44):
kind of going, "Hey, wait a second.
You've been learning on my dime.
Right.
created value on my dime."
So how do we partition out thisnew value that's been created
essentially by this, this new party,this new entity, this alien AI?
Where, is my share in that?
Yeah.
And it's happening.
(41:04):
You know, I call it the AIdiscount, which is happening.
And, you know, firms like-- if firmslike KPMG can go to its auditor and
say, like, "Hi. So I think that Ishould get a discount on my audit,"
which was highly reported it, it showsthat clients are definitely thinking.
So let's talk about the surplus.
So we now, let's say, with the helpof this technology, can produce
(41:27):
those outcomes using less hour, lesstime to produce the same output.
In the short run, let's say clientsdon't adapt, but ultimately, clients will
start asking the question that, "Hey,you know, you must be using some sort of
augmentation, so I should get a discount."
And so the surplus drains in a fewways, but there are two primary ways.
(41:47):
One will drain to the client, andit will drain to the client when
the work can be easily checkableand the stakes are relatively low.
And that work drains very quicklybecause the client's like, "You're
doing this, but I could get this modelto do this, and I can also test it."
So, you know, it's only when thestakes rise where the cost of
(42:11):
being wrong is wrong that they'llcome to you, but they'll still
argue with you on this discount.
The second leakage is now you have the newcost of the business is using a model, a
frontier model, a harness, or a controlplane to build-- get your answers out of.
Basically, you have to use-- either youuse your own platform or you use someone
(42:34):
else's platform to generate these answers.
And that platform owner, dependingon their position, is also going
to extract their rent out of thesurplus because they're like, "Hey,
you used these many tokens. Guesswhat? You owe me this much money."
Or we have a firm that I'm workingwith which sells something called
(42:54):
routines, which are loops thatthey sell to small businesses.
And they say, "We don't charge you anydifference on top of the token cost,
but every time you run this loop, we'regoing to charge you a small amount of
money." And that's again an example ofrent extraction coming out of to the
platform that you're losing money to.
(43:16):
And over the last two weeks, there hasbeen a lot of talk by the CEO of of, of
Palantir and lately Satya from Microsoft,which are saying that you need to become
sovereign in how you deal with your data,how you deal with the models, how you
host them, and you've got to do this.
(43:37):
And while both of them are arguingfrom their priors because they're both
saying like, "Listen, you could usethe Microsoft platform, or you could
use the Palantir platform," but they'rereally scared that platform capture
happens from the frontier models whowill then sort of like cut them out,
and that is going to be the challenge.
And that's where it leaks out, where therent Is slow to leak out is where the
(44:04):
output does not have a quick answer back.
Like, let's enter China.
I don't know.
Or let's launch this new product.
Will it be successful?
I don't know.
And or we'll find out in acouple of years, and the cost
of being wrong is really high.
In those situations, which is traditionalMBB strategy work and all of this, I
(44:26):
think it's going to be slower or let's sayaudit companies for Fortune five hundred.
I don't think that their rentwill go very quickly because
very few people can do that work.
And so your position in the marketplaceis going to depend or decide on
how much of that surplus you keep.
(44:46):
If you're competing in a place where youget answers back very quickly, the work
can be checked, the stakes are relativelylow, you're going to be in a world of pain
unless you figure out these locks that Italk about which can defend the business.
Otherwise you get competed ordeflated away in this environment
(45:07):
I, I'm loving your new vocabulary, man.
You're, you're introducingthis value drain and et cetera.
These locks, almost likea moat of the future.
A lock is smaller than a moat becausethe companies are smaller as well.
And, and to that, that point, maybe we'llfinish on this, so if delivery of a firm
moves from labor and th- that bottom base,the, the, the structure of the org like
(45:33):
that, to compute, computation essentially,
All right
how does a firm intentionallyproduce the next generation?
And remember when we did awebinar recently with Kyndryl and
Ismail Amela, he mentioned this.
He's like, going, "Imagine you havea review, a, a feedback review,
and it's with, it's with an agent,and you're treating this agent like
(45:55):
it's literally like a colleague."
Right
it came to mind when I was thinkingabout this, well, how, how do I both
create that new bench for the future?
'Cause actually, when you look at bookslike, for example, Stall Points, a book
about when companies get to these pointsof stall of revenue, what was the failure?
(46:15):
And of companies, itwas an internal reason.
It wasn't… 13% was actuallyregulatory change, sociopolitical
landscape, stuff like that.
was internal.
One of them was a, a talent benchshortfall, so no future talent that we
talked about, that academy, like the
(46:36):
Yeah.
story, that was able to carrythe new strategy for the future.
So you may come up with a new strategy.
We saw this in the shift fromanalog to digital in many companies.
For example, media companies, nobodyhad socio-media- social media literacy.
They weren't digital natives.
So you had this period that wasthis messy middle of transition.
(46:57):
What are we gonna see in an AIworld with that, particularly with
professional fir- service firms?
I think in the short run, it'sgoing to be really difficult for
smaller firms to build this benchout because to build a bench, you
essentially need to make an investment.
And when the investment was subsidizedby the client you were fine.
(47:21):
Like, I have lots of friends who run verylarge software companies or you know.
But the way that it worked was the juniorcame in, they error corrected, they
worked with-- They figured out how thesupervisor made the, the adjustments and
the error corrections, and ultimatelythey got harder and harder work, but
it was all paid for by the client.
Now we're talking about a world wherethat assumption is no longer true, and the
(47:46):
firm needs to make an intentional decisionwhether or not to invest in a bench.
And I think mechanicallyit is very possible.
I have friends of mine who are runninga simulation company which provide
simulation tools for companies to at leastget juniors started off, you know, from
zero to one and run through the simulator.
(48:09):
They allow you to quickly get started.
But I think, like, the next layer ontop is something called Shadowlands, and
I've, I've written quite a bit about this.
But essentially, mechanically, it is verypossible in this new world to do this.
There just comes a costto, to get it done.
And I think that's going to be thetrue decider as to whether or not
(48:33):
leadership today has the incentiveto do that if they're optimizing.
Because let's say that you are in apartnership, and you're gonna leave
the partnership in three years.
Are you going to invest indeveloping a brand-new set of talent?
And I read something recentlyabout Cognizant and some
(48:53):
of these large BPO firms.
They're actually making some headwaysinto this by making very large
investments into this pool today.
And I'll be writing about that soon.
But I can see firms intentionally sortof like betting with us because you have
to take margin, and especially for apublicly traded firm, that means you're
(49:16):
gonna make less money today, but hopefullyyou're gonna pay off it in the later day.
And this all comes down to incentivestructures as to whether firms are
gonna make this decision or not.
And fortunately the, the onus thereforefalls on the junior to try to figure
out new ways how to run apprenticeships.
(49:37):
But within the firm, it isa incentive discussion more
than anything else over here
it's a funny time and I, you know,I always think about that term, the
French term, après moi le déluge,which means after me the flood.
So you have partners in firmscoming towards retirement.
no incentive for them to go andre- re- reinvent, reshuffle,
(50:00):
Right
the entire organization.
So it's a really messy, messytime, and thank you for providing
shedding some light on it for people.
For people who wanna find out moreabout Framebreak, about your work,
where's the best place to find you?
Yeah, I think it's framebreak.com and Iwrite weekly over there, and a lot of the
(50:22):
concepts that we talked about over hereare discussed on the blog in great detail.
I sometimes write on LinkedIn as well,and those are the best places to connect.
And I look forward to sharing morein the coming weeks as I get closer
to releasing a lot of the thesiswork that I've been working on
And for those who enjoyed this,who learned from it, hope you have.
(50:46):
I'm also learning.
I told you I'm putting thesebuilding blocks in place, learning,
looking back over old, like, reallyearly books like that Design Rules
books really helped me understand.
More difficult to read.
I told Kim and Carliss Baldwin thatit was difficult to read, but really
worth the struggle to get through.
But the good news is, for meanyway, I'm gonna be collaborating
more with Usman in the future.
(51:07):
We might have a series, special seriescoming out or podcast, standalone
podcast, but we'll be collaboratingin some way in the future, and
really looking forward to that.
Finally, wanna thank our sponsor,Kyndryl, who run and reimagine
the technology for the world'sleading businesses and industries.
They work in aerospace, they workin healthcare, keep propping up
(51:27):
some of those massive industries,and it's a pleasure to have them
as sponsor of the Innovation Show.
You can also find the Kyndryl Institutewhere you'll find brilliant authors
like Usman, and many of the gueststhat we've had on the show before.
You can find that atKyndryl.com/institute.
It's been an absolute pleasurehaving you on the show.
Author of Framework and founderof High Output Ventures, Usman
(51:50):
Sheikh, thank you for joining us
Thanks for having me Aidan
Before we begin, I wanna thankour sponsor, Kyndryl, who run
(00:02):
and reimagine the technology thatdrives the world's infrastructure.
We're talking about healthcare, we'retalking about aerospace, and banking
amidst many, many other i- industries.
But also check out the KyndrylInstitute where you'll find
unbelievable thought leadership, manyof the guests we've had on the show.
You can find both at Kyndryl.com,Kyndryl.com/institute or .com where
(00:28):
you can find out more about Kyndryl
today's guest is a previous guest froma webinar we had recently with the
Kyndryl crew, and it's great to have him.
He's been writing lucidly about AIfor several years, and particularly
how it's gonna affect knowledge workand professional services companies.
was saying to him before we came onair, it was a little bit like that quote
(00:50):
where Wayne Gretzky was asked how doeshe always know where the puck is gonna
be, and he said he skates to wherethe puck is going, not where it is.
And that's a little bit what'shappened with our guest, because
he's been talking about this stuff.
Earlier on, people thought maybe,"Is this gonna happen? Is it real?"
Now they're listening, 'causethey've caught up with the puck.
He is the founder of High Output Ventures.
(01:12):
He's the author of a brilliant sitecalled Framebreak, Usman Sheikh.
Welcome back to the show
It's great to be on.
Looking forward to this conversation.
Always enjoy these where, like yousaid, the world of professional work is
changing, and there were signs last year.
But lots of people have talked about thedeath of consulting for a very long time.
(01:35):
and by no means do I think, like, there'sgoing to be the death of professional
work, but I definitely do think thatthe way that the firms were structured,
the way the work was delivered, isfundamentally going to change, and that's
what I'm most interested in writingabout and talking about these days
I was having breakfast recently with asenior leader, and I was saying to him
(01:56):
that there's two concurrent shifts ormultiple shifts going on, but the, with
regards to firm, was saying to him, I, wewere having breakfast, so I picked up a
salt cellar, and I was like, " the firmused to be like this, like an IBM." And
then I pointed to the gr- grains or theflakes of salt that were on his avocado.
And I was like, "See the flakes ofsalt there on you?" I said, "So the
(02:18):
firm, like IBM, had to break up tobe more agile and became modular.
then eventually by doing that, it invitedin lots of competitors, and then I poured
out some salt on the table, and he waskind of looking at me going, "What is
this guy gonna talk to me about?" AndI was like, "Now it's like grains." I
said, "But AI is ta- " I took some ofthe grains and I put them in my glass
(02:41):
of water, and it's like, it's dissolvingsome of those grains, and it's leaving
some, it's leaving only very few.
in there he went, he was looking atme and was like, " did that work?
Did you understand what I was sayingthere?" And he goes, "That's a
brilliant way of understanding it."And I was like, "Oh, yes." So I was
like, you've written then about thatstructure and new structures that
(03:02):
are needed, and I thought maybe wewould start with why was the structure
the structure in the first place?
And then we'll get ontowhere you see it going.
I think professional service firmssolved a-- the way that they were
structured solved a scarcity problem.
The firm's most valuable assetwas its senior leaders' judgments,
the senior people at the top.
(03:24):
But the number of hours in theday were limited and, you know,
they could only service a certainnumber of client requests.
In order to solve this problem, whatthey did was that they put partners
at the top and built layers ofmanagers and juniors beneath them so
that they could scale their judgmentacross a much wider array of clients
than any single one person could.
And I think that became theleverage unit that we see today, and
(03:49):
that's why the structure existed.
And more than just multiplying the outputof the senior leader, what it also did
was that the base of the firm was theprofit engine of these firms as well.
So the rate that the firm wouldcharge for the work versus the
rate which they paid the base, thedelta between that was the margin.
(04:12):
And the wider the base became,the more money the firm made.
And also the, the people at the base ofthe firm were actually very profitable
units of margin for the company, andtherefore it expanded its way outwards.
So two things came outof the pyramid so far.
One was leverage of the seniortalent's judgment, and the second
(04:34):
was the profit and the margin.
And the third thing, which I thinkis not talked about as much, is
that the pyramid also was a trainingground for the next generation of
leaders and signers at the company.
So when you have a wide base, you canfigure out who is good, who is doing the
work well, and through a series of errorcorrection, supervision through seniors,
(04:58):
these junior people ultimately becamethe people who could one day put their
name on a document and sign it as well.
And the beauty of this was that all ofthis training was paid by-- for the client
under the guise of the billable hour.
And what AI does is that it pulls allthree of these factors, the leverage,
(05:22):
the margin, and the formation of talentaltogether, which then puts a lot of
pressure back onto the firm becauseif they are to adapt, they're going
to have to figure out how to changea model that has worked remarkably
well for them for a very long time.
I love the idea of the pyramidand the base and the understanding
(05:44):
there that a training ground.
And then it opens up the question,and we'll… I- I go all, all
over the place, man, here.
You know
Right.
roll.
But I also fear that, we've seen thiswith all these firms, that they're
not hiring for that base anymore.
So I was thinking like, "Well, whatare they gonna do then?" obviously
(06:05):
training AI you know, there's somefirms bonusing people on using AI
or punishing them for not using AI.
But then on top of that, I've seenthis concept where if you have a senior
executive or a senior consultant fromone of those firms, they're creating
a digital twin version of that person.
Right.
And
Yeah
(06:25):
may be let go in the next few months oryears, but there's this digital twin.
So you have frozen that knowledge at thestart, at the top, and then maybe actually
filling in through outsor- sources ofinformation, new papers, et cetera, your
future work that you're about to publishin the future, stuff like that feeds it,
(06:46):
and then you have this, just a- a- a- a-an LLM perhaps that is owned by the firm.
Yeah
know that's n- not exactly theright language to put on this,
but that's the concept that we're
Yeah.
There as well.
Yeah.
So I think that youcorrectly outlined two parts.
One was the digital twins of seniorleaders, which we can park on one side.
(07:11):
And the other part, which I think appliesmore to the formation base, which is
that if you stopped hiring as manyjuniors into the firm, you don't create
a problem today in the market becauseyou can hire from other firms and other
people from the stock that exists today.
But maybe five years from now orseven years from now, the market
(07:32):
is going to face a glut of thattalent which does not exist.
So I think I really thought about the,the formation problem quite a bit, and
let's talk about the junior problem first.
So the first thing that I said was thatthe base of the the formation was paid
for by the client in the traditional modelbecause, you know, you learnt on the job.
(07:57):
Now, when the client, let's say, doesnot get billed for hours of work,
and they just get billed for, "Thisis the outcome that I would like,"
and you have to figure out how tobuild that for them, you take that
abstraction away from the junior base.
And now what the firm has to do ismake an intentional decision as to
(08:18):
whether they are going to invest intothe formation of talent in the future.
And I'll tell you from personalexperience, we have a IT services
company in the, the portfolio, andthis is a discussion we regularly
have because it just takes a longtime to grow that person, nurture
(08:40):
that person into that pers-- into thatperson who can sign in the future.
And the decision now is (08:44):
Shouldn't we just hire someone from the
market who is mid-tier and sort oflike pay them a little bit more,
but augment their output with AI?
And I think this is a veryintentional decision that someone
is gonna have to, to make.
Because you know, I have friends whoare working on building simulators
(09:06):
for junior talent at some of thesefirms that, okay, so if you're not
doing the work, there are simulators.
In one of our firms, what they make thejuniors do is that when the output comes
out from the LLM, the junior has to assessthe output, and then without seeing what
the senior or the system would recommend,and then they cross-triangulate,
(09:28):
and they're doing all of this work.
So all of this stuff is happeningright now, but who's paying
for all of that development?
It's the firm is payingfor that development.
And I think, like- It's the correctchoice for the firm to actually set
some budget to do this because theyset themselves up for durability of
(09:48):
competitive advantage in the future.
Because let's say that we do allof this and we skip the thing, five
years from now, we don't have a newstock of signers, that we can't afford
new signers or senior people in themarket anymore the firm unfortunately
may have harvested profits for fiveyears, but is going to find itself in a
(10:08):
really tough spot five years from now.
So, so I think that's a formation thing.
The core, if it is that intentionalityor talent formation becomes the
responsibility of the firm and the payoffis much further out, but it's perhaps the
correct decision to make in the short term
(10:29):
I see it exactly the same as a sportsteam and exactly the same as making
some bets in innovation that m- manydon't work out well, but a few may pay
for all the ones that didn't work out.
But back to the sports idea, r-because we're in the middle of the
World Cup, I'm gonna age our episodenow, but of course, it's about AI,
(10:50):
so it's gonna age very quickly.
But I recently said this aboutRonaldo, so Ronaldo, brilliant player,
absolute unbelievable respect to himfor playing so late in his career and
keep going and be able to perform.
But there's lo- huge questions overshould he have been playing, and for
me, what that means is he's preventingsomebody else from developing, and as
(11:12):
a result, and this is the choice ofultimately the coach or the management
team, preventing others from growing.
And I often think about that.
If you have a digital twin, say,of a senior consultant, and you're
relying on that versus to what yousaid, where I have somebody else going
through ultimately the struggle that'sneeded to actually learn, 'cause it's
(11:34):
through the struggle of learning.
We were talking about some of thepapers we read on AI before we came on.
I mean, it's a huge struggle to readthose papers, to get through, and actually
new terminology and terminology you'reunfamiliar with, but it's only through
the struggle that you actually learn.
Right
one of the things I wanted to say wasthis concept of the Chinese room that
(11:54):
we talk about in AI, where you actually,like, push a piece of paper under
the, the, f- under the, beneath the,
Yeah, a dork
and the AI does the work andpushes it back, and you've no idea
what happened be- after that whenit was pushed through the door.
That more and more, even if youare a junior coming into the bottom
(12:15):
of that base layer of the pyramid,many are using AI and not going
through the, the, the disciplineand the struggle of the learning.
And in there, you're like, don't know whoactually knows what anymore because all
you're ultimately selling is the output.
And it, it's, I, I think the governance ofthat is huge, and it can be missed so much
(12:37):
because you think you're doing the work,but it's actually the machine doing the
work all the time, and nobody's learning.
Yeah.
I wrote a post about this, and Icalled it fluent and substitutable,
which goes to the exact word that youjust did that you use the machine.
So I used an example of a firm where the,the junior was struggling to r-get to a
(13:00):
certain outcome, and they were relyingheavily on the machine to produce outputs.
And they thought that it's justsome iterations that I need to
do, a few more button presses,and I'll get close to that number.
And it doesn't work that way becauseyou've just converted the LLM into
a casino or a jackpot machine whereyou're just hoping that the next button
(13:22):
press is going to give you the answer.
And while it could happen, Idon't think anyone bases their
paying their rent on whether theslot is going to pay out or not.
And similarly my advice to that junior wasthat you have to do the work by hand to
understand what the machine is doing sothat you can actually grade the output.
(13:44):
And if you do those difficult reps, youactually understand what's happening
because if you don't, then everyone whocan use your same prompt or eventually
the machine will just absorb yourwork because if you don't know what's
happening, the machine is like, "We'reboth guessing over here," and the
machine gets more reps at it than youdo, the machine's gonna win at the end.
(14:08):
I was thinking of that movieWALL-E, you know, the, the I
think it was a Pixar movie, and
Yeah
for watching Pixar, but one of theanimation studios, WALL-E, this,
about this robot, and it's a worldwhere humans don't do very much and
we're all extremely obese 'cause we'veoutsourced everything to the machines.
And I was like, oh, well, that'skinda what a lot of people are doing.
(14:29):
Where I was talking to a, a teamrecently, and they were saying that if,
for example, throw this out there, thatyou're creating policies, huge policies
for huge organizations, but actuallyyou're almost just the person punching
in the data at the start, and it's themachine coming out with the policy.
(14:50):
So nobody actually understands the policy,and everybody's working towards those
policies, and you're kinda going, there'snuances in there that are gonna be missed.
There's gonna be somany thi- things missed.
There's also gonna belots of stuff spotted.
But in there, there's a huge threat.
So maybe so- some thoughts on thathappening, 'cause it, it is happening
all over the world, before we gointo talking about business process
(15:12):
optimization, outsourcing, offshoring
Yeah
of the professional firms use, andwhere that fits into all of this.
I think when we look at past moves suchas, like you mentioned, offshoring.
Offshoring essentially changedwhere the work got done.
It didn't change anything else.
It basically changed the jurisdiction,lower cost jurisdiction, and the
(15:36):
work was still done by people,but it was done somewhere else.
Workflow software, basically RPA and otheraspects, made predictable processes into
a structured format within encoded rules.
So it changed how you did the work,but within a certain structure.
(15:56):
And I think when I think about AI,what it does is that it takes this
reusable system which now can freezeitself from this constraint of rules
that were around and it changeswho is doing the work now, right?
So it doesn't produce… WhatI've talked about is it produced
judgment-shaped artifacts.
(16:17):
Like, you know, it's-- it doesn't havejudgment in it, but it's ju- shaped.
Like someone who doesn't know what agood brief looks like will probably
think AI has created a great brief,but someone who's created briefs
for a living will know, like, "Okay,this is not a bad brief, but it
made mistakes in these four areas."
And I think, like, that's how I seeAI being very different from the past
(16:40):
world, because in the old models, itwas still when the machine reached
or when the process reached a certainexclusion that it couldn't understand,
you always had to revert back to a senior.
But now that has expanded, and you needfar fewer people sort of like doing that,
and the machine starts everyone off notfrom zero, but from later in the cycle.
(17:03):
And I think that's fundamentally differentfrom past shifts because it has direct
implications on who's doing the workand how the work gets done compared.
Because past shifts, a lot oflegacy co-owners will say fortified
the pyramid more than anything.
They became bigger, like, you know,Accenture has like eight hundred
(17:26):
thousand people, a lot of them in,um, you know jurisdictions like
India and Philippines, et cetera.
But AI does not necessarily play outthat way because if you just use it as a
productivity enhancer, um, I think yourlong-term competitive advantage deflates
(17:49):
very quickly because everyone will usethe same tools to get to the same answer
You remind me of an old story ClaytonChristensen told in, I think it
was How Will You Measure Your Life.
He, he talked about the, he called it theship of Theseus, but the whole idea of the
ship of Theseus, for those who don't know,is, like, the ship gets replaced plank by
plank, and eventually you're kinda going,"There's no planks from the original ship
(18:13):
so wh- what, is it still the same ship?"he talked about, well, Dell outsourced
more and more of one of their machinesout to China, I think it was, with ASUS.
And ASUS then eventually, as Dellkinda went down the stairs from
their side, ASUS were going up thestaircase from their side, building
ta- capability more and more and more,and then actually becomes a competitor.
(18:37):
And we've seen thatwith BPOs, for example.
And actually, where you probably see itmore, and I haven't seen anybody really
call this out, is for example, of theEuropean or even a lot of the US companies
maybe use India, and now the knowledgework in India is absolutely amazing.
(19:00):
The people coming from India are amazing.
A lot of the new mortgages herein Ireland are actually Indian
people who have moved to Ireland.
So
Yeah
Kinda reversals of that, the sametype of thing of the outsourcing.
Yeah
in there is a huge threat.
So we talked about the threatof on-boarders, new starters
in some of these firms.
(19:21):
But then think about the other side.
If y- if you've a BPO out in India orwhat used to be an emerging market,
and I don't need that anymore,
Yeah
ecosystem, a huge house of cardsthat could come tumbling down.
Because somebody told me, you know,if you're working for one of these big
(19:42):
firms in a BPO over in India, you'reextremely well paid by local standards.
You're getting Ubers to work.
You're living in luxury locally.
But if that's gone now, not onlydo you fall, but so does the actual
house of cards that's dependent on you
Yeah.
And it's a major source of employmentwithin the, the country's base as well.
(20:06):
And a lot of their exports are directlylinked to this export-based revenue that
gets collected back into the country.
So it has large macro implicationsfor the country as well.
And there's this large narrativeshift of like, okay, well, we have
all this trained labor, like you said.
(20:27):
We're going to convert them intothese forward-deployed engineers
and, you know, suddenly they're goingto do this other new types of work.
And the reskilling narrative isdefinitely there, and there's also the
conversation that as the professionalservices work will get cheaper to produce,
(20:48):
then there will be a lot more work andtherefore, hopefully, a lot of these
people will find employment within that.
But I think there's two things in thatargument which I wanna click into.
One is like, invariably, if the answersget cheaper, you're gonna have more
people asking more answers for it.
(21:09):
So, so that's there.
But there are two things moving over here.
One is the demand for that unit ofwork, and the second is the human
labor cost to deliver that unitof work, however much is required.
And so if the demand rises faster than thehuman labor cost rises, then employment
(21:32):
will, will rise because you still needmore humans in order to fill the thing.
But if the human cost falls fasterthan the demand sort of like rises,
well, then the reverse happens and, youknow… And I don't think it's going to
be clear-cut across the entire industry.
I think serv-- certain service linesare going to still be heavy human labor
(21:55):
required, like high-stakes work is notgonna suddenly be handed over to the
machines because of a lot of things.
But low-stakes work, which is heavilycheckable, yes, I think a lot of
that work is going to be a threat.
And the call center industry is a verygood sort of like analogy of this.
(22:15):
There's a firm called Concentrix,and then I usually compare it to
a new firm called Sierra, whichprovides, again, voice agents.
One uses human labor to pick up the phone,and there's a certain cost towards it.
And Sierra is multiples inmarket cap value, and its
revenue is growing exponentially.
(22:36):
Like, I think they did a hundredmillion they reached a hundred
million in February, and they-Crossed two hundred million in June.
So it's basically the curve isexponential, and what they provide is call
center agents for regulated industries.
And there are no humans involvedin this transaction anymore.
(22:56):
And I think, like, that's where evenwhen the stakes are higher, you can
see the models moving into this.
This is just not picking up the phoneand booking a reservation for you.
This is someone calling up a bank andasking them a complex question, and that's
still being handled by an agent today.
Just goes to show you that it's notnecessary that increase in work is
(23:18):
necessarily going to lead an increasein employment because value capture
can either go to the person, to theplatform, to the client, or to the firm.
So… And that's the hard part inthis conversation that I'm finding.
Like, who captures all of thisgain that is going to occur now?
I, I really wanna talk about that,
(23:39):
'But just on that, what I thought about that before is it's almost
like, I call it this idea of musicalchairs, that game we used to play
as kids, where the chairs run out.
And you think about each shift we've gonethrough from digital to cloud to today,
there's been shifts, but I just don'tsee where there's enough chairs at the
(24:03):
table for people to be all employed whenAI is taking so much of the value and
Bye.
optimizing so much.
And as, as, as to, as what we talkedabout before, it's taken away that
bottom layer, so there's actually
Right
at the bottom layer g-starting, so there's no way
to get onto the layer at all.
Yeah
point there, the valuation of the firmis based on usually revenue per employee.
(24:27):
So one of the examples is Base44,this company that sold to Wix for
80 million had eight employees.
One shareholder, eight employees.
And you go, well, w- if that's thecase, and that's happening more and
more and more, and we're gonna see,as I said s- to you before, not two
pizza teams, but two slice companies,where there's enough just to feed
(24:50):
one person who owns those companies.
And it's a lot of the companiesyou've invested in, co-created.
Right
does the employment lie?
And, you know, people are kinda going,"Oh, well, I'm gonna get my kid to
become a carpenter or go into thetrades." You're kinda going, "Yeah,
but who's gonna buy the houses andemploy them?" So th- there's a whole
ecosystem here needs thinking about.
(25:13):
And the irony, I thought aboutthis, the irony is somebody in
government somewhere where he's kindagoing, "This is on my desk now,"
and they're typing it into an LLM.
Yeah.
Yeah, I think yesterday it, it was GoogleDeepMind founder Demis who wrote an
article on X in fact talking about howthis is this massive change happening
(25:37):
and how we need a collective responsein how to deal with the repercussions
that this is going to have on society.
Because I think the, the ju-juryis still out on is there gonna
be net employment increase?
How are we gonna distribute this?
I've heard arguments on both sidesof the, the thing, and usually
(25:59):
you're arguing your priors.
So if you want more employment, thenyou're gonna say, "Of course, more work
is gonna create more people," becauseif my margin unit is dependent on that
unit, then I do not want less becauseit creates a lot more problems for me.
But if I'm a technology-augmentedfirm which is using this, I'm
going to say the, the, the latter.
(26:22):
Ramp, in fact which is this card--expense card plus now accounting software
company, wrote a paper recently whereit found that high growth companies when
they adopted AI aggressively actuallysaw headcount rise within the firm.
But I think, like, what the papersays in one line was that this is
(26:43):
limited to high growth firms in sortof information technology sectors.
And so your, your concern is valid, andI think a lot of people are seeing that.
It's just very difficult because Ithink jobs are a bundle of tasks, and
what we spoke about earlier was thatif you unbundle the job, there are
certain parts of the job that sort ofget automated away, and the parts that
(27:09):
remain require you to either have highjudgment or the ability to have some
credential to stand behind the answer.
You might have a, a, a law background ora audit certificate or a CFA or a doctor.
I think that provides you position,but someone who doesn't have those
(27:29):
gets put in a very difficult spot.
Um, and like the first work that we weredoing, sort of now firms need to invest
in you to build those skills where it'snot a byproduct of how I do the work.
That in itself creates, um, tensionwithin the firm because do you optimize
for the short, mid, long, or altogether,or how do we do, or it's very uncertain.
(27:54):
I don't want to invest now.
Maybe I'll invest two years from now.
And these are all conversationsI'm having with portfolio
companies, and it's, it's not easy
You know, I saw this shift when I, Iwas a professional rugby player, as
you know, and I went into the workingworld in 2008, which was terrible
timing if you look at it that way.
(28:15):
But I was, I actually purposely went inat that stage because I was like, "Oh,
well, this is gonna be difficult, andI may as well go into an environment
that's difficult for everyone," andused the kind of perseverance and the,
the resilience that I'd built up fromsport to work in the working world.
But one of the places I first saw wasI saw salespeople really struggling be-
(28:38):
because before they'd actually workedas booking agents, or they were esser-
essentially just minding their clients,and now they have to actually sell, and
they'd actually forgotten how to sell.
And before we, we talk, 'cause Ireally wanna share the shape of the new
firms, you talk about five differentstructures, shapes that a, a firm can
have and the evolution of those shapes.
(28:59):
But the, the atrophy of those skillsif you don't use them, and one of
the things I see people, and thisis why I was talking about that
mental atrophy, that you may becomementally obese because you're not u-
actually using that muscle anymore.
There's a huge onus on people to becomeconstant learners or be lon- constant
learners, and I know it's one of themotivations behind Framebreaker through
(29:23):
your writing, my writing as well.
It's not only for the audience to sharethis stuff, but it's actually for me to
actually understand it, and going throughthe struggle of figuring out arguments,
does that work, does it land, my
Right.
my salt, metaphor at the start.
D- does it make sense?
And actually going through the processof that, thinking, sleeping on it,
(29:43):
thinking about it over several days.
And when people are using agents moreand more and LLMs more and more and
co-working more and more, the- allthey are are this person who shares the
information that they've just got fromthe LLM, even to a point of, "Upload
the PDF, give me the main points." like,when you read a book I, I used to, I
(30:04):
used to get this from people to go,"Oh, why don't you just use one of these
summary book- you know, like Blinkistor one of these tools to summarize books
for you instead of going through it?"
And I kind of go, Whenever I did that andthen I read the book, I used to do both,
I'd read it and then read the book, I'dpick out totally different highlights."
Yep
much of the book was so personalto m- my own experience, my prior
(30:27):
information that I'd sourced, that I'dbuild information totally different.
So if we all use the same agentsto produce the same results, to get
the same highlights, no innovationin there 'cause there's no spotting
anomalies and information that'sappealing to you or something that
might spark a metaphor for you, and Ijust think that that's a huge shame.
(30:50):
And I wondered what your advicewas for people who are… 'Cause
we're all doing it a little bit.
Yeah.
towards that.
What is your advice?
What, what are you seeing as a bestuse of the tools that are out there?
Yeah.
I like the, the way that you saidthat, you know, you were doing both.
I sometimes use that as afilter of whether I want to
(31:10):
read this completely or not.
Because also, to be honest, a lot of booksthese days, especially the newer books
these days, they fill a lot of pages, butthere are a few ideas that are stacked.
So you wanna figure out whetheryou wanna go through them or not.
But I completely agree with you thatthere are certain books where just the
(31:31):
highlights take all-- they abstracteverything away from the, the fabric
of how everything stitches together.
That if I just-- If you saw Alexander'sBusiness Model Canvas, and you
just saw that snapshot and yousay, "Okay, here are these boxes."
You put your firm on this, and then youtry to figure out w-- how to maneuver it.
I think, like, you know, it's, it's quiteshallow, but if you wanna understand
(31:55):
what is the, the, the core behindeach one of this, how do they each
interact, what are the different parts?
I don't think that you get that.
And that's how I, that's how Itend to read these days as well.
I think certain older books thatyou'll just read through because
there's just so much in those books.
But especially when I read newer books,I do this where I go to the parts which
(32:16):
is also most relevant to me so that I canpick that up, and then if that grabs me,
then I'm reading a lot more of the book.
And but, you know, I, I don't knowwhether the, the audience will see this,
but this Mike Kirsten had written thisbook called Project to Product, and
now his book just came out yesterdaycalled Output to Outcome, which is very
(32:38):
relevant to what we're talking aboutover here because we're moving from a
society which is was driven a lot byoutput, like, especially software firms.
Output was the code and everything, butnow we're talking about these outcomes.
But how do we reorient thefirm around the outcomes?
And he has a structure in thebook, and I read a synopsis of it.
Someone that I really respect r-recommended the book, and here is the
(33:01):
book, and I started reading it thismorning, and it was a great sort of, like
read so far because it connects with a lotof things which are relevant to me today.
So that's, that's my processas to how I'm doing things.
And it's becoming harder becausethere's just so much to read
these days that you have to figureout what's really important.
(33:21):
Seth Godin years ago saying that he, he'dput the summary… You know, the way you
see at the end of a chapter, it usuallygoes chapter takeaways, that he'd put
it at the start of the, of the chapter.
Yeah
through learning w- how the bestoptimization of learning is actually
you, if you have a, a concept of whatyou're gonna learn early, then you
(33:43):
kind of read to almost validate that.
But if you're led down the wrongpath, so this is what I was saying
about the AI could actually lead youdown the wrong path and go, "These
are the salient bits. This is what'srelevant," and then you read it.
You're missing therichness of it, you know?
And, and when I write, I, want to bringthe jour- the reader on a journey.
So actually, in a way,you're warming up their mind.
(34:06):
Like, the way I see it is the metaphoris creating, like, Velcro, and then
the idea, which sometimes is complex,is what you throw against the Velcro,
and it, it sticks to the metaphor, and
Yeah.
Yeah
d- you can't shortcut that.
Like, that has to go
Yeah.
that journey, and it's like, when I runa workshop, I, I kinda do the same thing.
(34:27):
You kinda gotta go through and bring themon a journey, not too long, but bring
them on a journey where, where it sticks.
But let's, let's talk about the, themove of what y- the way you s- you
phrased this, and maybe we'll qualifyand share some of what you mean by this.
So as leverage moves towardsan intelligence core,
(34:47):
Right
then different structures ofthe firm emerge, and there's
five that you identify.
So maybe let's talk about what, what doyou mean by leverage, what do you mean
by an intelligence core, and then these
Yeah
shapes that emerge
Right.
Let me go through it quickly.
We talked about leverage firstwhen we talked about the pyramid
and traditional structures, whichthe source of leverage was labor.
(35:10):
If you wanted to do more work, youwould add more labor, and that is how
this firm progressed or grew bigger.
Now, the intelligence coreis essentially taking the
learnings of how work gets done.
So let's make this real.
A client comes, and they want us toproduce ads for their e-commerce brand,
(35:34):
and we have to put them up on Facebookfor them and test it out for them.
So let's say that we tell theLLM to generate us, "Here's a
brand, here's all this context.
Generate some ads for us." So theLLM will push the ads out, and now
someone with knowledge of both thebrands and advertising will say like,
(35:55):
"Okay, let's edit this, let's trythis, and let's put this to work."
They'll put the ad. There'll be afeedback loop where the ad platform
will say, "Oh, this ad performed well,didn't perform well." Let's say that the
ad came back, it didn't perform well.
We made three tweaks to it,and then the ad did better.
Okay, so that's a simple loop thatI just created over there that, you
(36:18):
know, you could replace that with taxor anything else, or a legal contract.
Most of the times, you just log thewinning ad variation and say, "This
ad did better for every, And it'sjust logged in a library or a best
practice or a template or whatever.
An intelligence score is when the outputwas created by the machine, someone
(36:41):
with expertise was able to make errorcorrections to it, and then was able
to assess what happened with it or not.
And then they would feed all thoseerror corrections back into a
machine which learns all of this.
So, so for all the things that youwould discard, this is putting all
of that context back into the core.
(37:02):
And now the next time the machinesays, "Oh, here's a health brand which
is targeting the same demographic,"the next iteration that the machine
will produce will be of higher qualitybecause it has some learning feedback
loops from before that makes it better.
This is when you don't rely on aperson for a source of leverage, you
(37:23):
rely on the machine to produce thisartifact which abstracted that thing.
This also goes very close to your digitaltwin example that you used earlier
because, you know-- But there's a keydifferentiation I wanna make here.
Digital twin is a snapshot ofyour state of mind today, and that
(37:46):
decays at a really fast pace inthe world that we live in today.
The loop that I'm talking about isthis constant flow of information and
work that is happening at the firm,error corrections that are happening.
And let's say that we've never workedwith a shoe brand targeting women in
Europe, and today we're gonna do that,but we haven't created ads for it.
(38:08):
So someone is going to set that up first,do that, put it into the thing, and
the core becomes smarter and smarter.
Once this works really well, and you canstart to orient the entire firm using
this to ask it, "Here is the client. Weproduce this outcome. Can you produce this
(38:29):
for me?" And it produces output for them.
This loop creates, the feedbackloop happens, and the firm changes
structure because instead of addingan additional person to the team, you
can scale work depending on how muchdemand you can generate and how quickly
you can verify the work coming out.
And the firm could potentially morph froma pyramid to what I'm calling a ring.
(38:53):
And the ring would have theintelligence core at the center,
and then you would have this newstructure of people around it.
And in a nutshell, that'sthis transformation.
And I go through in, in detail forhow it could happen to various models.
But I think as an abstraction, thisshould provide you with an analogy of
what I mean by the leverage, intelligencecore, and how this new shape comes about.
(39:19):
It's funny you were saying thereabout the, have a snapshot of
someone's, essentially their mentalquiver at a certain point in time.
But what you see also is, you see peopleget to a certain level in organizations
and actually stop learning, and actuallythinking, "I, I have it now. I'm here
now. I, I, I've, I'm in port." Andyou're kind of going, well, there
(39:40):
is no safe port or safe haven with
Yeah.
anymore.
You have to consistently learn and finda way to learn, which is why learning
how to learn is, is so important.
And unlearn stuff as well because I thinkthat's where, where people have dug their
own grave in a way with AI, in that y-you gotta be the one who's constantly
(40:02):
provoking yourself to learn, and it,it's uncomfortable, and it's a pain.
Like, I would say, like, some ofthe papers we read are boring.
Like, they are tough academic reads.
They're tough to get through.
But it's only by doing that thatI actually can learn and, and
challenge your own thinking as well.
But you, you mentioned there,so the, the shape changes.
(40:22):
The, the gain then shifts.
And the, so you, you were saying earlieron about, okay, the, the, amount of w-
effort I put in versus what's output is,is, is my profit, essentially, my margin.
But there's a surplus created byAI, and then maybe the client's
(40:44):
kind of going, "Hey, wait a second.
You've been learning on my dime.
Right.
created value on my dime."
So how do we partition out thisnew value that's been created
essentially by this, this new party,this new entity, this alien AI?
Where, is my share in that?
Yeah.
And it's happening.
(41:04):
You know, I call it the AIdiscount, which is happening.
And, you know, firms like-- if firmslike KPMG can go to its auditor and
say, like, "Hi. So I think that Ishould get a discount on my audit,"
which was highly reported it, it showsthat clients are definitely thinking.
So let's talk about the surplus.
So we now, let's say, with the helpof this technology, can produce
(41:27):
those outcomes using less hour, lesstime to produce the same output.
In the short run, let's say clientsdon't adapt, but ultimately, clients will
start asking the question that, "Hey,you know, you must be using some sort of
augmentation, so I should get a discount."
And so the surplus drains in a fewways, but there are two primary ways.
(41:47):
One will drain to the client, andit will drain to the client when
the work can be easily checkableand the stakes are relatively low.
And that work drains very quicklybecause the client's like, "You're
doing this, but I could get this modelto do this, and I can also test it."
So, you know, it's only when thestakes rise where the cost of
(42:11):
being wrong is wrong that they'llcome to you, but they'll still
argue with you on this discount.
The second leakage is now you have the newcost of the business is using a model, a
frontier model, a harness, or a controlplane to build-- get your answers out of.
Basically, you have to use-- either youuse your own platform or you use someone
(42:34):
else's platform to generate these answers.
And that platform owner, dependingon their position, is also going
to extract their rent out of thesurplus because they're like, "Hey,
you used these many tokens. Guesswhat? You owe me this much money."
Or we have a firm that I'm workingwith which sells something called
(42:54):
routines, which are loops thatthey sell to small businesses.
And they say, "We don't charge you anydifference on top of the token cost,
but every time you run this loop, we'regoing to charge you a small amount of
money." And that's again an example ofrent extraction coming out of to the
platform that you're losing money to.
(43:16):
And over the last two weeks, there hasbeen a lot of talk by the CEO of of, of
Palantir and lately Satya from Microsoft,which are saying that you need to become
sovereign in how you deal with your data,how you deal with the models, how you
host them, and you've got to do this.
(43:37):
And while both of them are arguingfrom their priors because they're both
saying like, "Listen, you could usethe Microsoft platform, or you could
use the Palantir platform," but they'rereally scared that platform capture
happens from the frontier models whowill then sort of like cut them out,
and that is going to be the challenge.
And that's where it leaks out, where therent Is slow to leak out is where the
(44:04):
output does not have a quick answer back.
Like, let's enter China.
I don't know.
Or let's launch this new product.
Will it be successful?
I don't know.
And or we'll find out in acouple of years, and the cost
of being wrong is really high.
In those situations, which is traditionalMBB strategy work and all of this, I
(44:26):
think it's going to be slower or let's sayaudit companies for Fortune five hundred.
I don't think that their rentwill go very quickly because
very few people can do that work.
And so your position in the marketplaceis going to depend or decide on
how much of that surplus you keep.
(44:46):
If you're competing in a place where youget answers back very quickly, the work
can be checked, the stakes are relativelylow, you're going to be in a world of pain
unless you figure out these locks that Italk about which can defend the business.
Otherwise you get competed ordeflated away in this environment
(45:07):
I, I'm loving your new vocabulary, man.
You're, you're introducingthis value drain and et cetera.
These locks, almost likea moat of the future.
A lock is smaller than a moat becausethe companies are smaller as well.
And, and to that, that point, maybe we'llfinish on this, so if delivery of a firm
moves from labor and th- that bottom base,the, the, the structure of the org like
(45:33):
that, to compute, computation essentially,
All right
how does a firm intentionallyproduce the next generation?
And remember when we did awebinar recently with Kyndryl and
Ismail Amela, he mentioned this.
He's like, going, "Imagine you havea review, a, a feedback review,
and it's with, it's with an agent,and you're treating this agent like
(45:55):
it's literally like a colleague."
Right
it came to mind when I was thinkingabout this, well, how, how do I both
create that new bench for the future?
'Cause actually, when you look at bookslike, for example, Stall Points, a book
about when companies get to these pointsof stall of revenue, what was the failure?
(46:15):
And of companies, itwas an internal reason.
It wasn't… 13% was actuallyregulatory change, sociopolitical
landscape, stuff like that.
was internal.
One of them was a, a talent benchshortfall, so no future talent that we
talked about, that academy, like the
(46:36):
Yeah.
story, that was able to carrythe new strategy for the future.
So you may come up with a new strategy.
We saw this in the shift fromanalog to digital in many companies.
For example, media companies, nobodyhad socio-media- social media literacy.
They weren't digital natives.
So you had this period that wasthis messy middle of transition.
(46:57):
What are we gonna see in an AIworld with that, particularly with
professional fir- service firms?
I think in the short run, it'sgoing to be really difficult for
smaller firms to build this benchout because to build a bench, you
essentially need to make an investment.
And when the investment was subsidizedby the client you were fine.
(47:21):
Like, I have lots of friends who run verylarge software companies or you know.
But the way that it worked was the juniorcame in, they error corrected, they
worked with-- They figured out how thesupervisor made the, the adjustments and
the error corrections, and ultimatelythey got harder and harder work, but
it was all paid for by the client.
Now we're talking about a world wherethat assumption is no longer true, and the
(47:46):
firm needs to make an intentional decisionwhether or not to invest in a bench.
And I think mechanicallyit is very possible.
I have friends of mine who are runninga simulation company which provide
simulation tools for companies to at leastget juniors started off, you know, from
zero to one and run through the simulator.
(48:09):
They allow you to quickly get started.
But I think, like, the next layer ontop is something called Shadowlands, and
I've, I've written quite a bit about this.
But essentially, mechanically, it is verypossible in this new world to do this.
There just comes a costto, to get it done.
And I think that's going to be thetrue decider as to whether or not
(48:33):
leadership today has the incentiveto do that if they're optimizing.
Because let's say that you are in apartnership, and you're gonna leave
the partnership in three years.
Are you going to invest indeveloping a brand-new set of talent?
And I read something recentlyabout Cognizant and some
(48:53):
of these large BPO firms.
They're actually making some headwaysinto this by making very large
investments into this pool today.
And I'll be writing about that soon.
But I can see firms intentionally sortof like betting with us because you have
to take margin, and especially for apublicly traded firm, that means you're
(49:16):
gonna make less money today, but hopefullyyou're gonna pay off it in the later day.
And this all comes down to incentivestructures as to whether firms are
gonna make this decision or not.
And fortunately the, the onus thereforefalls on the junior to try to figure
out new ways how to run apprenticeships.
(49:37):
But within the firm, it isa incentive discussion more
than anything else over here
it's a funny time and I, you know,I always think about that term, the
French term, après moi le déluge,which means after me the flood.
So you have partners in firmscoming towards retirement.
no incentive for them to go andre- re- reinvent, reshuffle,
(50:00):
Right
the entire organization.
So it's a really messy, messytime, and thank you for providing
shedding some light on it for people.
For people who wanna find out moreabout Framebreak, about your work,
where's the best place to find you?
Yeah, I think it's framebreak.com and Iwrite weekly over there, and a lot of the
(50:22):
concepts that we talked about over hereare discussed on the blog in great detail.
I sometimes write on LinkedIn as well,and those are the best places to connect.
And I look forward to sharing morein the coming weeks as I get closer
to releasing a lot of the thesiswork that I've been working on
And for those who enjoyed this,who learned from it, hope you have.
(50:46):
I'm also learning.
I told you I'm putting thesebuilding blocks in place, learning,
looking back over old, like, reallyearly books like that Design Rules
books really helped me understand.
More difficult to read.
I told Kim and Carliss Baldwin thatit was difficult to read, but really
worth the struggle to get through.
But the good news is, for meanyway, I'm gonna be collaborating
more with Usman in the future.
(51:07):
We might have a series, special seriescoming out or podcast, standalone
podcast, but we'll be collaboratingin some way in the future, and
really looking forward to that.
Finally, wanna thank our sponsor,Kyndryl, who run and reimagine
the technology for the world'sleading businesses and industries.
They work in aerospace, they workin healthcare, keep propping up
(51:27):
some of those massive industries,and it's a pleasure to have them
as sponsor of the Innovation Show.
You can also find the Kyndryl Institutewhere you'll find brilliant authors
like Usman, and many of the gueststhat we've had on the show before.
You can find that atKyndryl.com/institute.
It's been an absolute pleasurehaving you on the show.
Author of Framework and founderof High Output Ventures, Usman
(51:50):
Sheikh, thank you for joining us
Thanks for having me Aidan
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by Audiochuck Media Company.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com