What does it mean for culture when ‘intelligence is on tap’?

First of two episodes going deep on how AI is going to impact work – and consequently workplace culture.

This week’s discussion is with Alexia Cambon from Microsoft.

Alexia is Head of Research on Copilot & Future of Work. Last month her team released the Work Trend Index Annual Report. It’s one of the most important pieces of insight into how our jobs will change. Their previous reports have been interesting going deep into how people are experimenting with AI but this year’s is different. It articulates a version of work that most of us aren’t yet ready for.

Links mentioned in the discussion:

Transcript

Bruce Daisley (00:00)
Alexia, thank you so much for joining me again. I wonder if you could kick off by introducing who you are and what you do.

Alexia (00:07)
Sure, yeah, very happy to be here, Bruce. My name’s Alexa. I’m head of research for a team here at Microsoft called the Co-Pilot and Future of Work team. And what we do essentially is try to understand how AI is impacting how people work and what that means for companies.

Bruce Daisley (00:25)
And I guess your big reports, I’m not sure if you did one last year, I was looking out for it, but the work trend annual report, certainly there’s definitely one for 2025, but it’s just come out and it seems a real step change for me from what you were saying in the last report. The last report, really, certainly the last one I vividly remember is 2023, which we were talking about. dealing with these work demons of like the broken meeting and the blank page. And yeah, that’s right. And it was like this really vivid articulation of where AI was in the moment. And what you’ve published right now seems far more visionary and far more articulating something that is both.

Alexia (00:58)
blank page.

Bruce Daisley (01:17)
exciting and potentially a huge change from where you are. So I’d love you to just talk about what you think the big findings of the new report are.

Alexia (01:25)
Sure, yeah, and thanks for that lovely feedback. I think this year really felt like a different research process for us because beyond just doing our annual survey of 31,000 people and beyond looking at the telemetry data, which is obviously some of the most exciting data to look through, we also just did a bunch of interviews with academics and scientists and economists to just really understand how they were thinking about the future of work. And that’s really the research question we asked ourselves was what does the future of the firm look like and what does that mean for employees? ⁓ So yeah, top level findings.

I think first of all, it’s clear from the survey data, 82 % of leaders said 2025 is a pivotal year to rethink strategy and operations. So we knew from the outset that this would be a year where a lot of important decisions were going to be made. The three findings we outlined in the report, the first is that intelligence is now on tap.

And that’s a pretty unique, big concept when you think about it in the way of companies run on intelligence, they run on expertise. And up until this point in time, we’ve only ever really been able to access expertise and intelligence through humans. And that has often been very costly, especially if you want to access that intelligence quickly. And now for the first time, we’ll be able to access that in addition to humans outside of humans.

So that was the first thing we really explored. We understood that when you can suddenly access expertise in that way, almost like buying electricity, like a commodity, that has a ripple effect on how the organization is structured. And that was the second big finding, is that human AI teams are gonna become the norm.

And as we start to think about AI joining the team and working with humans and having this new digital workforce, what does that mean for how we organize work and for the org chart in particular?

And then the third final finding was really about the implications for the individual employee and how if we now all of a sudden have human AI teams, what will be expected of the employee and what will the employees work be made up of? And we believe that every employee will be required to become what we call an agent boss, essentially the CEO of their own little agentic startup, and that they will be working very much with their own teams of agents that they will delegate work to.

Bruce Daisley (03:59)
was trying to articulate to someone on a Sunday afternoon. So like you can imagine it’s already an unreceptive audience. But I was trying to articulate to someone the findings of this and you know, the notion of you talk about a frontier firm and like how these things are different. And I was really struggling to explain to someone how work might feel different.

with this environment, which I mean, you’ve just described three revolutionary changes and I was sort of struggling to articulate it. I wonder how you sort of set about trying to describe the differences. I was trying with, in very tactical ways to just describe the like, the look and feel of work, working with agents on a day-to-day basis. And I think possibly because this feels, even though we’re using AI at the moment.

It feels a few years away that we’re in a team with agents. How do you, how do you try and bring it to life to people? What does their working week look like?

Alexia (05:06)
Yeah, well, great question. And for that, we are extremely blessed that we have access to real objective data as to what the shape of the workday looks like. And when I say objective data, mean survey data is very much how do people feel? What do they think?

But telemetry data, which is all of the signals that the N365 products emit that we collect as part of our metadata at the aggregate anonymity, that is what is actually happening. So that’s our best way of really determining what does work look like. And when we did a deep dive this year, it was very clear that work, knowledge work anyway, information work is incredibly sloppy and very inefficient and very painful to a lot of people. So some of the things we saw, for example, is we’re interrupted.

every two minutes and that’s just within the entry 65 suite so that’s not taking into account if you have whatsapp open or spotify open just within emails teams meetings you are interrupted every two minutes

By the way, my favorite part of this data is getting real life flashes of inspiration that really caused me to want to dive deep into this data. So I’ll give you an example. I run commute to work and I had this pretty early meeting one day. It was at 7 a.m.

call, so I ran in at some ungodly hour and five minutes before the call it got cancelled and I was incredibly frustrated and upset and so I asked my team to look into whether or not we’re experiencing a last-minute culture at work and so one of the things we looked at was in the 10 minutes before a meeting what is the degree of editing, frantic editing happening in the PowerPoint deck.

And we saw a 122 % boost of frantic editing in the PowerPoint deck in the 10 minutes before a meeting compared to the three hours before. So our hypothesis was proven accurate that there is this crazy last minute culture happening.

Bruce Daisley (07:05)
love it.

Because that’s one of the bits of data actually that’s really caught people’s attention and made people think. And that was brought about by a genuine insight of what you were experiencing.

Alexia (07:21)
Exactly. Yeah. And so I think, you know, these signals that show us and I could go on and on, we saw that,

over 50 % of meetings are scheduled ad hoc. We saw that just all these signals around how disrupted and last minute things are happening really point us towards a world in which it takes work to do work. And I think you asked me, what do we think the future looks like? I think a lot of that pain and a lot of that work that we’re doing just to do work will be automated away.

taken on by agents. And so you can imagine a world in which the last minute editing of the PowerPoint is done by an agent or the scheduling of the meeting is done by an agent. Or rather than you logging into your Outlook in the morning and looking at your emails, you actually just go to your agent and ask your agent to tell you what are the most important items that I need to know from my inbox. So essentially I have this theory that a lot of this…

very painful, inefficient work that is required to do work is going to be done by agents in the future.

Bruce Daisley (08:35)
It posed a really interesting question for me. spoke to someone who’d created a new product that was about sort of reinventing, it’s a product called Roam. it’s, you know, it’s like one of those new inventions that is based on an interesting hypothesis. And these interesting hypotheses is that in the office of the future, we will think about calendar zero as our aspiration, rather than inbox it. And the very

His whole point was, is because I guess when agentic processes are dealing with the weekly catch up or the weekly status meeting, it’s only things that humans need to intervene in that are going to be of consequence. And they might be spontaneous rather than just in the weekly cadence. Right. And it, it’s such a sort of disruptive thought that was really taken by it. And especially as I read your stuff, the following week,

And I wonder what your perspective on that is. Do you think that the idea that we’ve got this cadence of 25 hours a week of meetings that we were constantly checking our inbox, is that going to be something we escape from or is actually a requirement to check in with our agents going to require us being present far more? How do you visualize those specific details?

Alexia (09:54)
mmhm

Yeah, so couple of thoughts. think the first right now, all of us are paying a consumption tax as information workers on all the work we do, which is it’s become so easy to communicate that we are just constantly landed with information that we have to consume and process. I think that consumption tax will shift over to the agent. The agent will be consuming the bulk of communications for us, but the tax we will pay instead is a delegation tax, where we’ll be delegating a lot of the work to the agent to do that on our

I think that would be very liberating, but it doesn’t mean we won’t be paying a tax. It would just be a different tax. In terms of calendar zero and meetings, we have some good science on this already. So we did a large scale randomized control trial of AI in the workplace. It was a 12 month study where we looked at the differences in work patterns between co-pilot users and non-co-pilot users. And that gave us the first real hints as to how the workday is changing based on the use of AI.

And we saw that across the 58 companies that participated in this study, 6,000 subjects that time spent on email was consistently reduced. That was the most consistent finding we had, was just less time in inbox. And if we’re already seeing that just in the first 12 months of Copilot being released, I feel quite confident that that will only continue.

And that potentially, rather than starting our day in Outlook, we’ll start our day with our agents telling us what we need to know from email. The meetings thing is more interesting and there’s much less consensus on that. When we looked at what was happening within meetings across this 12 month period, there was no consensus. Some companies had increases in time spending meetings, some had decreases in time spending meetings. And that needs me to believe that it’s not really about AI as a silver bullet, it’s about your meetings culture.

And are you really making a concerted effort to change the meetings culture by using AI as opposed to just expecting AI to change that by default? I will say on the calendar zero piece specifically, it’s interesting. I’m seeing this trend of us moving from, and our chief scientist, Jamie Teevan talks about this, moving from documents to dialogues where up until this point meetings have become about

information exchange, we just log on to meetings to exchange a bunch of information. And I think I see an era where meetings instead become about value creation. And I’ll give you a real life example from a company we worked with on this study. They had given co-pilot to a group of white paper authors whose whole job was to break down the chemistry components every time a new product was released. And the way they used to do this, these white paper authors, is each white paper author would write up

their section, you know, each specialist would write up the

chemistry components, they would send it off to the next specialist, et cetera, et cetera, et cetera, a very long, arduous, asynchronous process. What they started doing with Copilot is that everyone would get onto a call and they would write the paper out loud. And by that, mean each specialist would say, this is what’s gonna live in my section. Together, they would discuss the abstract and the conclusion, and then they’d hand the transcripts to Copilot, and Copilot would write the first draft of the white paper.

And in that instance, you could see a world in which actually you’re doing more meetings because creation happens in those meetings, not outside.

Bruce Daisley (13:31)
The whole thing poses such a fundamental question to me in you’ve talked about value creation, but once we reach a situation where information is where intelligence is on tap, once we reach a situation where intelligence on tap, then one of the big differentiators that we’ve always had in our own competitive markets to some extent is undermined. And maybe it’s undermined by

the average quality of work in the sector going up. But we no longer have the edge of believing that we’ve got more talented people working here than our competitors have. And it just strikes me then that there might be value creation done in other things. know, from my perspective, I wonder if culture is going to be a bigger differentiator. Or I wonder if you’ve talked about meeting culture, actually, as a small nuance, that could be an interesting one. What’s your take on

What do you think are going to be the differentiators, the value creators in this world where superintelligence is on tap?

Alexia (14:40)
Everyone always asked me like, what skills should my children build? And I think this is a slight tweak on that question that this brings to mind is what skills will AI rareify?

That to me is the more interesting question because if you now have an agentic workforce working alongside a human workforce and that agentic workforce has inherited all of the expertise that there is to inherit, whether that be legal expertise or marketing expertise or finance expertise, and that is no longer your differentiator as a human employee, what are the things that AI can’t do that you can? And that to me will be the value differentiator for humans. And I think it’s everything to do with

And I know this is a cliched boring answer, but it is everything to do with the more human skill sets that are unique to humans. And by that, I don’t necessarily mean empathy because we know AI can often present as more empathetic than certain humans. There are loads of medical studies that have been done on that. But I’m talking about, for example, my most senior clients will not want to talk to an agent about a really sensitive business deal. Right. Doesn’t matter how adept the agent is.

that relationship needs to be protected by having a human to human connection. know, speaking of medical diagnoses, maybe I don’t want to get a medical diagnosis from an agentic doctor, right? Maybe that again, feels like I want to be sat in a room with someone looking me in the eye, who I feel knows what I’m going through, because only a human can really know what I’m going through, right? An agent is not a human.

So storytelling, relationships, business transformation, strategy, all of those much ⁓ more complex, less tangible aspects of the work experience, I think will be more reserved for humans moving forward.

Bruce Daisley (16:36)
I’d love to go back to that. That notion that work right now is interrupted 200. It’s such a powerful data point, 275 times a day, because it’s, I mean, so it’s such a colossal number. I remember vividly, like a decade ago, someone had a data point that we pulled our phones out of our pockets 50 times or out of our purses 50 times a day. And that seemed like a big thing. Like 275 seems like a…

even higher magnitude of, of interruption. Do you think even before we get into wrestling with these agentic disruptions, we could put our business in a better place by thinking about that experience of work right now? Do we need to the groundwork of thinking about where work is today before we throw forward into maybe three years where, these things are a reality? I’d love your take on that.

not least how many years into the future are these realities.

Alexia (17:40)
Yeah, I mean, if well.

For one, what my team and I started thinking about was trying to explain this world to someone in the 1950s and imagine the look on their faces. And I kept thinking about a worker sitting in the secretarial pool or sitting in their office, writing out or typing out memos and imagining their boss jumping into their office every two minutes. That would be the equivalent, right? Because there was no email, there was no IM. The equivalent would be someone physically

your office every two minutes. And when you think about it in that way, it does seem absolutely gaudy that this is what’s happening to our brains within the digital ecosystem. So while I always think it’s worthwhile pausing and assessing the work culture that we’re in, if we know that agents are coming anyway, and we’ll be able to help with a whole load of this, you know, my gut feel is, you know, that is what work is going to look like moving forward. So optimize and design for that.

rather than going through the very arduous, hard process of changing habits. And certainly, you know, there is good etiquette and there are good cultural norms and standards we should put in place, but we should be realistic about how effective they are when the current work culture is so insidiously filled with these disruptions. There are some that work really well, I think. I mean, for me, I think I shared this with you before, Bruce. Like every time I’m about to send an email, I ask myself if this email cost me 50 pounds to send, would I send it?

right, because we’ve eliminated all communication costs so we have to reinsert those costs in somehow. ⁓ But I am very much looking forward to the day where I don’t need to physically remind myself of that every time I write an email because, you my agent will be working with your agent, Bruce, to ensure our communications are only ever really about the high-value stuff.

Bruce Daisley (19:37)
I wonder if you read the AI 2027 predictions by Daniel Kokotajlo and what your thoughts on that were.

Alexia (19:47)
I don’t think you can be an AI researcher and not have read that, Bruce. Well, I mean, I’m an incredibly data-driven person, right? That’s my job. So I think I always look to the science we have available to us to draw what seems like.

Bruce Daisley (19:51)
Okay, what was your take on that?

Alexia (20:05)
fairly firm conclusions, whilst also realizing that every futurist who ever made a prediction, 95 % of the predictions they made turned out to be wrong. So yeah, think the 2027 report was very much predicting and projecting out scenarios that…

they believe could occur. I think I always look more short term based on the science that we’re seeing now. What do we expect could happen? Yeah, as the data is indicating. And so for example, the human agent teams finding in the report.

That very much came from the study that Harvard did with Procter and Gamble that really looked at how does AI join the team and what happens when you give an employee access to AI compared to a whole team that doesn’t have access to AI and can they perform at the same levels. So we very much drew inferences from the science available to us.

Bruce Daisley (21:07)
The P &G study was fantastic, but I guess the one thing that that study, and it was truly eyeopening for an illustration of where we currently are, know, people who get carried away with where we’re going to be, was an illustration of the power of where we currently are. But I guess the big difference that you’re articulating here is that the agents are going to truly have agency. They’re going to have some sort of degree of autonomy. They’re going to be doing stuff for us. And it begs the question,

And when we all got communication on our mobile devices or a generation of work, not saying I was in that generation, but I doubt was, you know, 15, 20 years ago, when we got that, we all thought it was going to transform our experience for work. And actually what it’s done is, is tethered us to work more than ever before. And so do you believe the autonomy of these agents is going to be liberating for us or is it going to be?

Alexia (21:45)
Hahaha!

Bruce Daisley (22:07)
Deja vu and the requirements of these agents is actually going to be even more demanding of our attention.

Alexia (22:16)
I think that’s up to us to decide. and I also, I think it’s very easy to look at the negative aspects of any new technological innovation and completely take for granted the positive aspects. And so, you you mentioned mobile, now being able to access work, you know, on your mobile.

and to be able to eliminate the very arduous costs associated with only ever being able to access your work in the physical office, in the physical, you know, it even a laptop, right? Or they called the stationary PCs that we couldn’t take home with us, right? For many people, I think especially people who weren’t really taken under consideration when the workplace was originally designed, I’m thinking working parents, I’m thinking, you young women,

would have been a hugely positive development to be able to access work from home.

⁓ But I think we are bound to always think about the negative aspects of these things. ⁓ When it comes to agents, it’s definitely a really interesting question how much the lines are continuing to blur across personal and work. And if, for example, you now have a personal agent that sits across both your work and your personal life, can you more effectively integrate your schedules ⁓ so that you can manage both ⁓ more?

Bruce Daisley (23:42)
interesting.

Alexia (23:43)
effectively

and I think

That again will be down to us humans to determine whether or not we use it for good or for bad. And at the end of the day, it’s always going to be the human oversight that requires us to say work is taking over too much. And now I am now putting a boundary in place that I refuse to take calls after 6 p.m. or I refuse to sacrifice my 45 minute run in the middle of the day for work. Like an agent can’t do that for you. That is the human motivation that

you cannot automate away. Conor Grennan talks about this a lot and I love this as an example, which is if AI is a treadmill you can go by the treadmill and you can you know put out the training plan of your dreams to train for a marathon but you have to get on the treadmill and you have to run the miles and I think it’s going to be the same with agents. We’re still going to need the human motivational factor to be able to use them the right way.

Bruce Daisley (24:44)
Final question, because I know we’re out of time. the one thing that, the reason why I mentioned the AI 2027 Daniel Kokotajlo is that because I think the thing that comes across from that is that none of us are ready. None of us have corporate five-year plans right now are not ready for what’s going to happen. And the reason why I was so moved by your work is that like this, it feels a step change from the last time I read a report. it’s broadly what I would say is…

Alexia (25:02)
ha!

Bruce Daisley (25:13)
you’re not preparing properly for what’s coming or, you know, embrace, embrace this because it’s going to be completely disruptive. And that’s why I find it interesting that when we talk about meeting culture and the notion that a lot of organizations have got 30 hours a week of meetings and you say, well, that’s down to the company’s philosophy. I just wonder if unless you’re willing to embrace some degree of much bigger degree of disruption, you are going to be one of these frontier firms.

Unless you’re embracing that your calendar is going to need to look different, the way you work is going to need to look different. Almost that’s the tell that indicates that you’re not set up for the change that’s coming. Change seems wonderful in abstract and once you described the specifics of it, people often flinch. And I just wonder if that’s the one thing that maybe people need to think about.

Alexia (26:08)
for sure.

Change readiness and willingness to embrace what’s coming. Yeah. And I think when we looked at respondents in the survey that could realistically be working for frontier firms, the percentage was tiny, tiny. know, the people who actually met the criteria outlined like full scale deployment of AI, high AI maturity, awareness and understanding of agents. You know, very few people actually met the criteria. And I think it’s not even looking at this from just a company level. It’s looking at it from an economic standpoint.

societal level and we’ve just published a paper where we put out the question you know organizational theory is organized around the idea that we have two inputs to productivity which is human labor and capital that augments that human labor.

And we’ve been categorizing AI as capital because it looks and feels like a software. But if you actually look at all of the traditional criteria that we use to define capital, it doesn’t meet any of them. It self-improves. It scales massively. It’s elastic. And as a result, our theory is actually, what if we now had a whole new function of production where we now have human labor, capital, and digital labor?

and you need to organize for that, right? Like if you think about…

Capital we’ve got IT to look after capital to look after your laptops and your software. We have human resources to look after human labor What do we have to look after digital labor? Like right now it probably lives in IT, right? But should it that’s a huge question So even just from like a structural structural point of view it raises real questions about how we’re gonna manage and measure this new type of workforce and you know the P&G study my good friend

Karim Lekhani who ran that study, we’ve been talking about this idea of do we need an intelligence resources department? If we have a human resources department, do we need the equivalent for agents? I’m not saying we’ll treat agents like humans at all, but we’ll need to have a way to measure and monitor them. We’ll need to have a way to manage them. And yeah, that’s just one tiny example of how much change and disruption is likely coming.

Bruce Daisley (28:25)
Leave me with one thing. What are you reading? What are you listening to? Where are you staying? How are you staying atuned to this astonishingly fast moving field?

Alexia (28:38)
Well, so I have my AI stuff and then I have my non-AI stuff. So I’ll give you one from each if that works. So Mustafa Saliman, The Coming Wave, great book. Obviously Mustafa works at Microsoft, full disclaimer, but I want to very much stay in tune with how he’s thinking about the world because especially on that whole work personal front, I think he’s thinking very interesting thoughts. And then on the non-AI specific stuff, and there’s a point to this.

Bruce Daisley (28:47)
Perfect.

Alexia (29:08)
I’m reading Katherine Switzer’s ⁓ Marathon Woman memoir. Katherine Switzer, for those who don’t know, she was the first woman to officially run the Boston Marathon and I’ve been thinking a load about women in this space and not…

not having women be left behind in the AI race. There are a lot of male voices out there in the AI world and Catherine Switzer was one of the leading people to bring the marathon to the Olympics, the women’s marathon to the Olympics. So I’ve been trying to learn from how she did that to see how we can make sure women are represented in the AI space as well.

Bruce Daisley (29:48)
Alexia, I thought the report this year was like a huge breakthrough in terms of articulating where we’re going and not flinching from actually describing things that maybe some people might be afraid to discuss. So I’m so honored that you took the time out to talk to me. Thank you so much.

Alexia (30:08)
Aw, thanks for having me, Bruce. Absolutely great conversation, loved it.