Rebecca Hinds (00:00)

Welcome to Intelligence Real and Imagined from the Work AI Institute at Glean.

‍

Rebecca Hinds (00:05)

This is the show where we sort through what's real, what's hype, and what actually works with AI at work.

‍

Rebecca Hinds (00:13)

I'm your host, Rebecca Hinds, and I lead our Work AI Institute here at Glean.

‍

Rebecca Hinds (00:18)

And throughout this series, I'll be joined by my co-host, Bob Sutton.

‍

Rebecca Hinds (00:22)

Bob is a Stanford Professor Emeritus.

‍

Rebecca Hinds (00:24)

He's an organizational psychologist.

‍

Rebecca Hinds (00:26)

He's a bestselling author, one of the founding members of the Work AI Institute, and he's a longtime collaborator and friend.

‍

Rebecca Hinds (00:34)

This is our very first episode.

‍

Rebecca Hinds (00:36)

Thank you so much for joining us.

‍

Rebecca Hinds (00:37)

We're kicking things off with the AI Transformation 100.

‍

Rebecca Hinds (00:42)

This is our flagship report from the Work AI Institute.

‍

Rebecca Hinds (00:46)

Bob and I collected insights from more than 100 leaders, researchers, and builders, and distilled them into a hundred ideas that just might change how you work with AI.

‍

Rebecca Hinds (00:59)

And today we're joined by Arvind Jain, CEO and Founder of Glean.

‍

Rebecca Hinds (01:05)

We're talking about what happens when AI starts changing the fundamentals of how we work, how it reshapes expertise and the balance between experts and novices, how change spreads top down and bottom up, how to move fast without breaking things, and how to turn AI from a flashy pilot into something that actually works.

‍

Rebecca Hinds (01:30)

Let's dive in.

‍

Rebecca Hinds (01:35)

So, Arvind, it seems like every leader right now is saying they want to transform their organizations with AI.

‍

Rebecca Hinds (01:41)

It separates the organizations that are effectively driving change versus the ones that aren't.

‍

Arvind Jain (01:48)

Yeah.

‍

Arvind Jain (01:48)

One of the things that we have noticed that, you know, that drives success is leaders who are focused on business outcomes and thinking about how AI can actually impact those core metrics that they have in the business, as opposed to just like, hey, let's use more and more AI in the enterprise.

‍

Arvind Jain (02:07)

And also like leaders who are sort of fundamentally thinking about AI as a core part of their strategy.

‍

Rebecca Hinds (02:15)

Bob, that's definitely something we've seen in our conversations with leaders recently.

‍

Rebecca Hinds (02:20)

It needs to be a commitment.

‍

Rebecca Hinds (02:22)

What else have you seen in terms of the leaders that effectively are driving change versus not?

‍

Bob Sutton (02:26)

Well, the main thing I've seen is spending the last few months with you doing all sorts of interviews to develop this list of 100 sort of ways that AI might transform work.

‍

Bob Sutton (02:36)

But one of the things that really has, I think, struck both of us is this notion that when leaders see things in terms of black and white on all expertise, all novices, all centralization, all decentralization, whatever, those are the ones I worry about, the ones that seem to be doing the best are the ones who are navigating these tensions and trade offs in an evidence based and curious way.

‍

Bob Sutton (03:02)

I would describe it as I think.

‍

Rebecca Hinds (03:04)

Those tensions are something we've seen time and time again.

‍

Rebecca Hinds (03:07)

One that we detail in our report is between novices and experts.

‍

Rebecca Hinds (03:13)

In many ways, AI is changing who has expertise in organizations, organizations changing who has the monopoly over expertise.

‍

Rebecca Hinds (03:20)

Bob, there's this fascinating debate right now as to whether AI makes generalists more powerful in organizations or specialists.

‍

Rebecca Hinds (03:29)

Where do you stand?

‍

Bob Sutton (03:30)

Well, I hate to be an academic and say it depends, but I mean, just for example, it depends more where you are in the creative process.

‍

Bob Sutton (03:41)

Experts are really, really great when you know you're going down the right path and you just want to have them implement what has been proven and always done.

‍

Bob Sutton (03:52)

But the problem is that when people are experts in the wrong things and they're not seeing it with an open, curious eye, what you get is simply replicating the mistakes of the past by people who are powerful.

‍

Bob Sutton (04:08)

And we see actually AI can enable both of those things to happen.

‍

Arvind Jain (04:12)

So to me it's a lot of a matter of when and where.

‍

Bob Sutton (04:16)

Rather than experts or novices, who do you pick?

‍

Bob Sutton  (04:20)

Arvid, it sounds like you've thought about this some too.

‍

Arvind Jain (04:24)

Yeah, I mean, I think the AI is bringing that capability, that information, that knowledge at everybody's fingertips.

‍

Arvind Jain (04:36)

And so I think it does up level quite a few people with capabilities that they didn't have before.

‍

Arvind Jain (04:43)

And so I feel like it is an equalizer in some ways and it's enabling false score behind to sprint forward.

‍

Rebecca Hinds (04:53)

Yes.

‍

Rebecca Hinds (04:54)

So the second tension we see in this report is between centralization and decentralization.

‍

Rebecca Hinds (04:59)

This question of where should AI live in your organization?

‍

Rebecca Hinds (05:02)

Should it be centralized in a center of excellence or should it be pushed more to individual teams to manage and oversee cases?

‍

Rebecca Hinds (05:10)

It's a hybrid.

‍

Rebecca Hinds (05:12)

Bob, how do leaders think about this trade off between centralization?

‍

Bob Sutton (05:16)

Well, the whole tension.

‍

Bob Sutton (05:18)

People have been arguing whether centralization is best and there's people who want authority and want to be in charge and then there's the people who want freedom and everybody should be able to do whatever they want.

‍

Bob Sutton (05:27)

And that argument has always really annoyed me because it is a classic case of, well, it depends on what you're talking about.

‍

Bob Sutton (05:35)

If, if you want to have risk mitigation or you want customers, for example, to have a consistent experience.

‍

Bob Sutton (05:42)

I'm sorry, you need standardization and you need centralization to make sure it's sort of the same.

‍

Bob Sutton (05:48)

They don't let you do whatever you want.

‍

Bob Sutton (05:50)

When you want a McDonald's just for example.

‍

Bob Sutton (05:53)

But to the extent that you want to have variation in creativity, that's where you push the authority down the organization and you encourage people to try a whole bunch of different weird things, hopefully in a way that you won't mess things up.

‍

Bob Sutton (06:09)

My own employer, Stanford University, we've got this wonderful thing, my favorite current employee benefit, the AI playground.

‍

Bob Sutton (06:15)

It's got virtually every major AI tool and they keep upgrading it.

‍

Bob Sutton (06:21)

They encourage all employees to use it and figure out way and students, anybody else to use it in their jobs to try to figure out how to use it.

‍

Bob Sutton (06:31)

Whether you're a low-level staff member or you're running a huge research lab, you can figure out how to mess around with it.

‍

Bob Sutton (06:38)

So a lot of it depends on essentially whether you want to have standardization or creativity.

‍

Bob Sutton (06:42)

And there is a tension between the two.

‍

Rebecca Hinds (06:45)

And I think it's fascinating.

‍

Rebecca Hinds (06:46)

Just as the org structure, you sort of need to think about that balance in technology.

‍

Rebecca Hinds (06:50)

You do as well.

‍

Rebecca Hinds (06:51)

Right.

‍

Rebecca Hinds (06:51)

You need a central source to understand that everyone's following standards, it's secure, responsible, but you also need to give teams the flexibility to be able to develop those workflows and customize them to their, their ways.

‍

Bob Sutton (07:06)

I mean, so one of the mottos we talk about, which I stole from my friend Diego Rodriguez in the report, is he asked this question, where's your place for failing?

‍

Bob Sutton (07:14)

And any place that you're going to learn by failing without really doing too much damage, I think that's where you want some decentralization.

‍

Bob Sutton (07:22)

But I, I don't want it in a coronary care unit at Stanford, for example.

‍

Bob Sutton (07:25)

I want a little standardization.

‍

Arvind Jain (07:27)

Yeah.

‍

Arvind Jain (07:28)

I think there's also one more, you know, think here, which is, well, ultimately, you know, AI, you know, has better be a core part of every business process in every single department in your enterprise.

‍

Arvind Jain (07:41)

If, if it is not, then you're not leveraging the technology as you should.

‍

Arvind Jain (07:45)

So it has to be.

‍

Arvind Jain (07:46)

And, and, and to make that happen, you know, it has to be decentralized in some way.

‍

Arvind Jain (07:49)

Like, you know, every department has to sort of think, you know, AI native.

‍

Rebecca Hinds (07:52)

Yes.

‍

Arvind Jain (07:53)

And, but the problem is that right now, first everybody's busy and they don't have time to figure out what AI can do for them.

‍

Arvind Jain (08:02)

There's a lot of education that needs to happen.

‍

Arvind Jainn (08:06)

It does make sense in my opinion.

‍

Arvind Jainn (08:07)

When you have a centralized team, it can actually serve the purpose of ensuring that progress is happening.

‍

Arvind Jain (08:14)

They're not responsible for anything other than making AI work in your enterprise.

‍

Arvind Jain (08:19)

That focus helps then not only bring the consistency, as you said, of using AI the right way across all the different departments with the right security, with the right governance, but also they just push it, they drive education, they help every team understand the power of AI.

‍

Arvind Jain (08:37)

And in fact, if they don't make progress, they're going to be trouble.

‍

Arvind Jain (08:40)

So I think that actually was also important because all the other functions they got their core function and AI is just only one, an additional thing.

‍

Arvind Jain (08:49)

But for the centralized AI team, they're fully measured on success with AI.

‍

Arvind Jain (08:54)

So it helps to have that.

‍

Arvind Jain (08:55)

And it may be a temporary phenomena.

‍

Arvind Jain (08:57)

I think ultimately I think AI will become part of our basic existence.

‍

Arvind Jain (09:01)

So those teams may not actually necessarily be long lasting.

‍

Bob Sutton (09:07)

So I mean, one structural solution, which I'm thinking of other mature things.

‍

Bob Sutton (09:13)

Let's just spread.

‍

Bob Sutton (09:14)

Accountants talk about some financial.

‍

Bob Sutton (09:17)

We take them for granted.

‍

Bob Sutton (09:18)

There's always some finance person in every team.

‍

Bob Sutton (09:22)

I think we'll know that AI has been really spread when there's an AI person along with the finance person, along with the marketing person, along with the HR person.

‍

Bob Sutton (09:34)

And then we'll see it matrixed in a lot of big companies like Procter and Gamble.

‍

Arvind Jain (09:38)

In fact, I suspect they're already doing it knowing how Procter & Gamble works.

‍

Bob Sutton (09:42)

But to me that's a sign.

‍

Bob Sutton (09:43)

It's a more mature sort of discipline.

‍

Bob Sutton (09:46)

And in the case of let's just take finance or accounting, at Procter & Gamble there is a central accounting group, but then there's a whole bunch of people spread out across the different businesses too.

‍

Bob Sutton (09:58)

So to me that's the kind of sign that we'll see it's a mature field, if you will.

‍

Rebecca Hinds (10:02)

Yeah.

‍

Rebecca Hinds (10:02)

And Arvind, one of the steps you've taken at Glean is to create this new role that we call an AI outcomes manager.

‍

Rebecca Hinds (10:09)

Talk to us about why you decided this role was important and some of the impact you're seeing.

‍

Arvind Jain (10:15)

Yeah, well, so Glean is an enterprise AI platform and we give our customers this ability to really go and with their business people, not even with their technology people, they can go and build agents and automate some of their business processes.

‍

Arvind Jain (10:31)

And from our side we thought we built a really easy-to-use product that is intuitive, and I can talk in natural language and build agents and automate all this work that I had to do manually before.

‍

Arvind Jain (10:46)

And so we thought that this is going to be a quick success and as soon as you bring the tool to an enterprise, everybody's going to adopt it and all these business processes are going to be done they will build agents.

‍

Arvind Jain (11:01)

But what we saw was it was actually hard, it didn't happen.

‍

Arvind Jain (11:04)

The tool is easy to use, but people actually don't understand what AI can really do for them.

‍

Arvind Jain (11:10)

People don't even know how to prompt it, how to ask it, what the right capabilities are.

‍

Arvind Jain (11:15)

And it was clear that there was a need for education.

‍

Arvind Jain (11:19)

This team that we built, the AI outcome managers, their job is to work with customers, understand what are the business processes that are the most time-consuming, the most costly ones, and then bring and share of those business processes.

‍

Arvind Jain (11:39)

What are the things that AI can actually today do a good job automating?

‍

Arvind Jain (11:43)

They're bringing just that education of what AI can do for you.

‍

Arvind Jain (11:47)

And they also bring shared learnings from our entire customer base to each new company that we go and work at.

‍

Arvind Jain(11:53)

And that has actually driven a lot of adoption and success for our agents.

‍

Arvind Jain (11:58)

So mostly it's the role of education.

‍

Rebecca Hinds (12:00)

Yes, yes.

‍

Rebecca Hinds (12:02)

It's fascinating, right?

‍

Rebecca Hinds (12:03)

Because we see with centralization, decentralization, often there's these creation of new roles, either informal roles or formal roles.

‍

Rebecca Hinds (12:12)

Bob, we've encountered some situations where leaders are grappling with whether to create a new role or whether this new thing is a skill that should be vested in multiple people's roles.

‍

Rebecca Hinds (12:26)

We've heard of prompt engineering as one example.

‍

Rebecca Hinds (12:28)

Should prompt engineer be a role or should it be a capability that's vested, embedded in multiple roles?

‍

Rebecca Hinds (12:35)

Bob, how do leaders decide?

‍

Bob Sutton (12:37)

It's interesting because it's one of those questions that every day I change my mind about it slightly, but I guess where I'm at.

‍

Bob Sutton (12:44)

It's funny because I just reread our report the last two days from beginning to end, mostly working little pieces.

‍

Bob Sutton (12:50)

And the thing that really strikes me is I actually don't care exactly whether it's infused in everybody or their specific role.

‍

Bob Sutton (12:59)

What I care about is people are actually using it on the job every day, versus just doing window dressing.

‍

Bob Sutton (13:09)

And the window dressing can be everything from somebody who seems to have a visible prompt engineering or head of AI, blah blah, blah, role and doesn't have any power, versus somebody who really has a lot of power and is influencing everybody.

‍

Bob Sutton (13:24)

So I guess that's where I'm at.

‍

Bob Sutton (13:25)

If there's a lot of different ways to skin this cat.

‍

Bob Sutton (13:28)

But what I worry about is whether it's window dressing or real use.

‍

Bob Sutton (13:32)

So if a company buys your product but doesn't use it, it's kind of cool to make the sale, but it's not going to spread if everybody actually uses it.

‍

Bob Sutton (13:41)

That's the other thing is that they'll buy stuff and then not use it, which is the classic thing that happens with all known products and not just software.

‍

Arvind Jain (13:49)

Yeah, it's actually like, that's a really good point.

‍

Arvind Jain (13:52)

The like, you know, are you becoming an expert at prompt engineering and use, you know, driving the models to do some work for you, versus well, somebody you know in your company did that for you.

‍

Arvind Jain (14:05)

Now are you actually making use of AI on a continuous basis?

‍

Arvind Jain (14:09)

I think it's enough.

‍

Arvind Jain (14:09)

It's enough as you said, like, you know, to even just like not become, you know, that expert at prompting, but just like, you know, making AI part of your day-to-day use.

‍

Arvind Jain(14:18)

I think you know that automatically at some point, like, you know, it'll help everybody even develop those prompt engineering skills, the more you use the product.

‍

Arvind Jain (14:26)

You don't have to start being creative on day one.

‍

Arvind Jain (14:30)

You use first and then you learn how to improve it.

‍

Bob Sutton (14:34)

Ya it’s the Ethan Wallack, you get 10 hours.

‍

Arvind Jain (14:35)

After 10 hours, you can do something.

‍

Rebecca Hinds (14:40)

Speaking of Ethan Wallack, another topic of conversation right now is around flattening the org chart.

‍

Rebecca Hinds (14:46)

Does AI enable you to have fewer layers within the organization?

‍

Rebecca Hinds (14:51)

Bob, we spoke to one of our good friends and colleagues, Michael arena as part of this report and he drew the distinction between heads up versus heads down work as a key decision factor in helping you decide how much you can flatten to what extent.

‍

Rebecca Hinds (15:07)

What did you learn from Michael?

‍

Bob Sutton (15:09)

Well, Michael, so Michael, he is not a normal business executive.

‍

Bob Sutton (15:14)

He's been head of talent of AWS and General Motors is our big freaking companies.

‍

Bob Sutton (15:19)

And also he's a world-class academic researcher all at once.

‍

Bob Sutton (15:23)

I don't even know how he does this stuff.

‍

Bob Sutton (15:25)

And he's got his team, has some great data.

‍

Bob Sutton (15:29)

And it's interesting when you think of NVIDIA that shows it always doesn't apply.

‍

Bob Sutton (15:33)

When people get more than about 10 people reporting to them, they start getting overloaded, exhausted and the people who work for them don't do a good job because there's bottlenecks and all this sort of stuff and they're not getting enough personal development.

‍

Bob Sutton (15:45)

So that was the general effect of his.

‍

Bob Sutton (15:47)

But why don't you talk a little bit more about heads up versus heads down, because that's one of the limiting factors that he talks about.

‍

Rebecca Hinds (15:55)

And I think it's a framework, it's not a rule.

‍

Rebecca Hinds (15:58)

But he describes heads-up work as you're building relationships with people, you're communicating, you're collaborating, versus heads-down work.

‍

Rebecca Hinds (16:07)

You're very much independent work.

‍

Rebecca Hinds (16:09)

And his argument is the More heads down work is happening in an organization, the more you can afford to flatten.

‍

Rebecca Hinds (16:16)

If you start to flatten when you're surrounded by a lot of heads-up work, you're going to start to overwhelm the people who need to be engaging in those relationship-building activities.

‍

Rebecca Hinds (16:27)

Arvind, I'm curious because you've seen multiple large companies and thought about organizational structure, you're thinking about the org structure at Glean.

‍

Rebecca Hinds (16:36)

Do you think about the layers within organizations and the steepness differently for Glean now versus say when you were co-founding Rubrik?

‍

Arvind Jain (16:46)

Yeah, well I think first AI does give us that ability to do some of this heads-up work, like you said, in a much more effective fashion.

‍

Arvind Jain (17:00)

For example, if you are trying to understand what each one of your team members is doing, what work they've done, what help do they need?

‍

Arvind Jain (17:10)

Before, you had to actually spend a lot of time with each one of those team members to try to understand the scope of their work.

‍

Arvind Jain (17:18)

Now you can actually have AI go and really create those great summaries of what every one of your team members is up to.

‍

Arvind Jain (17:26)

And so, understanding the work of your team, finding opportunities to help them wherever they need help, a lot of this stuff can actually come and become a really good aid for you.

‍

Arvind Jain (17:39)

So I do feel like there's an opportunity for everyone to be able to increase the span, not just the team sizes, but also the type of work they can do.

‍

Arvind Jain (17:56)

And it will as a result help us sort of get to more flatter organizations.

‍

Arvind Jain (18:04)

Now I think it should be clear that I think everybody desires the organizations to be more flat.

‍

Arvind Jain (18:11)

Obviously it's more cost-effective but also it sort of makes communication better in some ways.

‍

Arvind Jain (18:20)

Like you don't have like, you know, the trickle down from the top to the bottom.

‍

Arvind Jain (18:23)

Like you know, the information loss reduces like you know, the more the, the organization is flat.

‍

Arvind Jain (18:28)

So I think it's a great, it's like, it's one of the greatest gift, you know, that AI is going to actually bring to organizations.

‍

Arvind Jain (18:34)

So, so we are at Glean similarly thinking about it.

‍

Arvind Jain (18:37)

Like, you know, I guess mostly it is for me it is about just fighting with my team members like who are trying to, who are trying to always, you know, get more people or bring more, build more layers.

‍

Arvind Jain (18:47)

And like, you know, now I, you know, and now I can actually ask them this question that how are we managing the team right now?

‍

Arvind Jain (18:54)

And why can't we actually, why can't we manage a few more before we create one more layer?

‍

Bob Sutton (19:03)

So this is something that it's just a huge topic right now.

‍

Bob Sutton (19:06)

But I'm thinking about what the Jensen Wong situation, where he's got 65 direct reports, how he does it and everybody keeps thinking about that.

‍

Bob Sutton (19:17)

So I ask a friend of mine, his name's Pete Newell, he's now a defense contractor, but he had 4,000 troops reporting to him in Iraq.

‍

Bob Sutton (19:25)

So he's an old combat leader.

‍

Bob Sutton (19:28)

And it was interesting.

‍

Bob Sutton (19:29)

Both he and Jensen say the same thing, which is the mission is the boss.

‍

Bob Sutton (19:36)

And so I said, what does that mean?

‍

Bob Sutton (19:38)

He said, well, you can have a really flat organization.

‍

Bob Sutton (19:42)

He said, all the people who were reporting to me, I have any control over them.

‍

Bob Sutton (19:46)

They're out in the field and these doing these little skirmishes and stuff.

‍

Bob Sutton (19:49)

But when they understand my intent and what the intent and mission of the organization is, I don't have to watch them that closely.

‍

Bob Sutton (19:58)

And I think that's something that it's interesting because Jensen brings that in the debate and I think that's in the flat versus tall discussion.

‍

Bob Sutton (20:06)

I think it needs to be brought in more because everybody knows what they're supposed to do, then you don't have to have a lot of one-on-ones with them.

‍

Bob Sutton (20:15)

And you're also not constantly correcting mistakes they make because they know what to do.

‍

Bob Sutton (20:20)

So I'm really thinking a lot about that is that if people know what they're supposed to do, you don't have to manage them that much.

‍

Bob Sutton (20:25)

And AI can help with that.

‍

Bob Sutton (20:27)

As you just said so well

‍

Rebecca Hinds (20:28)

And Bob, we know from previous work, it's about half of people can't tell you what they're responsible for in terms of goals.

‍

Rebecca Hinds (20:35)

They can't tell you the organizational goals.

‍

Rebecca Hinds (20:38)

And without that clarity, you absolutely need more structure.

‍

Rebecca Hinds (20:43)

Another tension that we uncovered in the report in our conversations was around speed and fast and slow.

‍

Rebecca Hinds (20:50)

How you know, there's a lot of pressure right now to move quickly with AI.

‍

Rebecca Hinds (20:54)

When do you move quickly, versus when do you slow down?

‍

Rebecca Hinds (20:57)

Bob, this is something you've studied in.

‍

Bob Sutton (20:59)

So many years too much.

‍

Bob Sutton (21:01)

The best I can do now for organizations is to use a race car analogy, which is that if you just go pedal to the metal the whole time, you're just going to smash into the wall.

‍

Bob Sutton (21:14)

But the way to go fastest down the straightaway is to know when to brake so you can make it out of the turn, know when to pit stop and well, when the car blows up, it's time to leave the racetrack.

‍

Bob Sutton (21:23)

And in some ways I think that that analogy works in the companies I've seen Over the years, not just in AI, but the ones that know when to sort of slow down and to figure out what's going on and what needs to be fixed.

‍

Bob Sutton (21:35)

But when they know what to do, then it's, let's shut up and put pedal to the metal and not argue about what we're going to do.

‍

Bob Sutton (21:45)

That rhythm, I think, applies to almost all human endeavors.

‍

Bob Sutton (21:50)

We've all been in organizations that go too fast and ones that go too slow.

‍

Bob Sutton (21:54)

The worst ones are where, after a decision is made, people don't believe it's a decision.

‍

Bob Sutton (21:58)

They still argue the decision; they try to undermine the decision.

‍

Bob Sutton (22:01)

That's not what you want.

‍

Bob Sutton (22:04)

But in organizations that work great, people just have a sense of when to go fast and when to go slow.

‍

Bob Sutton (22:09)

In the report, we have all sorts of empirical evidence that, well, that's the hallmark of great companies is they know when to slow down and fix things or figure out what to do.

‍

Bob Sutton (22:18)

But when it's time to go, everybody goes.

‍

Rebecca Hinds (22:21)

What's your viewpoint on when, if any, situations in terms of when pilots work versus when they tend not to work effectively?

‍

Arvind Jain (22:30)

Yeah, if you look at the last two years, enterprises, you know, they did a lot of POCs, lot of pilots, and many times they didn't even have a clear reason for why they were doing it.

‍

Arvind Jain (22:43)

It was more like, well, this looks like a cool capability.

‍

Arvind Jain (22:47)

Let's go and see how well it works in my enterprise.

‍

Arvind Jain (22:50)

And you're not yet tied them to, well, what are my key business priorities?

‍

Arvind Jain (22:54)

What are the metrics that I'm trying to actually move?

‍

Arvind Jain (22:58)

And therefore, should I actually be more careful in terms of what projects I should be really first starting?

‍

Arvind Jain (23:05)

So I think what I feel is the right approach is first think about not doing 100 projects at the same time.

‍

Arvind Jain (23:17)

Pick the five most important ones across your enterprise, and they have to be associated with what are your top business priorities?

‍

Arvind Jain (23:31)

And then instead of having the mindset of that, hey, I'm just trying this experiment or a POC, you start with the intent of that I'm actually going to build something in production.

‍

Arvind Jain (23:42)

Sometimes when you start these POCs and you don't have your teams really bought in, for example, many of these AI capabilities that you're trying to bring, they're hungry for data.

‍

Arvind Jain (23:54)

They need to actually connect with your internal systems, understand your business, and then try to actually do something, you know, with AI to like, automate those business processes.

‍

Arvind Jain (24:02)

But if you have these POCs and you're not really bought in, like, sometimes you've seen customers will say, that, hey, well let's actually, let's do this POC.

‍

Arvind Jain (24:11)

We don't have to connect it with our real systems, you know, let's just connect it with some test data and see what happens.

‍

Arvind Jain (24:16)

And that's sort of all signs of that.

‍

Arvind Jain (24:17)

Like, look, you know, you haven't really have, you know, you don't have the right teams bought in, you don't have the business behind you and you know, the POC ends and you sort of are left with like, well, like, you know, should we do it, should we not do it?

‍

Arvind Jain (24:30)

So those are some of the things which I would say like, sort of like try to stay away from and like start again with what are the top five things, you know, for our company and like get that support from the beginning.

‍

Arvind Jain (24:41)

Your security team should actually know that this is an important project.

‍

Arvind Jain (24:45)

It's not a POC.

‍

Arvind Jain (24:46)

This is something that you're rolling out in production so that we can sort of go smoothly and work with real data and solve problems.

‍

Bob Sutton (24:55)

I think this is a really interesting point.

‍

Bob Sutton (24:57)

I'm just thinking of every failed management movement I can think of, at least ones that faded.

‍

Bob Sutton (25:05)

Design thinking, which I was part of Agile, which I've been on the edge of.

‍

Bob Sutton (25:09)

And I don't mean agile software development.

‍

Bob Sutton (25:10)

Agile got spread to everything at one point.

‍

Bob Sutton (25:14)

And you end up with this situation where people think that they're adopting something by doing a whole bunch of cute little experiments and they experiment for experiment sake or they prototype for, but they don't actually ever want to implement it, they just want to do it to show they're cool and have fun.

‍

Bob Sutton (25:33)

So that is, it's just because Rebecca actually helped with this book.

‍

Bob Sutton (25:37)

We wrote a book on scaling about 10 years ago, Huggy Rao and I, and this is a definition of bad scaling is when you just sort of dabble and aren't thinking, what's the step to spread it across the enterprise.

‍

Bob Sutton (25:49)

Or worse yet, you spread something across the enterprise which isn't strategic, which wastes a bunch of resources and screws up the company.

‍

Bob Sutton (25:55)

So that's even worse.

‍

Bob Sutton (25:56)

So I think you're absolutely spot on because creativity for creativity sake is actually useless.

‍

Bob Sutton (26:02)

It's kind of fun, but if you don't have a path to implementation, you're messing things up.

‍

Bob Sutton (26:07)

So I really like your attitude is rare actually.

‍

Rebecca Hinds (26:11)

And I also think there's a component of labor of love.

‍

Rebecca Hinds (26:13)

The fact that if you fully invest in this, sometimes it's called the IKEA effect.

‍

Rebecca Hinds (26:18)

Truly build something.

‍

Rebecca Hinds (26:19)

If you're in the trenches, you're going to be more committed to it down the road anyway.

‍

Rebecca Hinds (26:23)

And we saw that in multiple points throughout the interviews.

‍

Rebecca Hinds (26:25)

Right.

‍

Bob Sutton (26:26)

So one of the most interesting interviews we did with a guy I've been teaching with for years, Perry Claibon.

‍

Bob Sutton (26:30)

But he's a real entrepreneur.

‍

Bob Sutton (26:31)

He's founded companies and everything and more than 100 classes have been founded in his companies, in his classes at Stanford.

‍

Bob Sutton (26:38)

And he said one of the downsides of AI for coming up with prototypes to start a company is he says it's just too easy to develop a good idea so you don't struggle enough for it.

‍

Bob Sutton (26:49)

So then when you test it, you don't struggle enough to have it actually succeed.

‍

Bob Sutton (26:54)

So he's having this kind of easy come easy go effect even when it's a great idea.

‍

Arvind Jain (26:59)

So it's funny that AI can make things too easy, not enough struggle.

‍

Rebecca Hinds (27:03)

And then the last tension we unpack is top-down versus bottom-up change.

‍

Rebecca Hinds (27:08)

We know that absolutely you need top-down change in organizations.

‍

Rebecca Hinds (27:12)

You need policy principles, but you also need bott change, and everyone needs to be using the technology to see real change.

‍

Rebecca Hinds (27:20)

Bob, how do leaders think about this balance between top-down change and bottom-up change?

‍

Bob Sutton (27:25)

Well, there's always arguments about what's better and I hate to be a boring academic, but to say yes when we see system-wide changes happen in organizations.

‍

Bob Sutton (27:35)

Well, first of all, it's the leader's job to make the decision, got to reinforce it, you got to provide the resources and you have to stop doing things so people can focus.

‍

Bob Sutton (27:45)

That's all great.

‍

Bob Sutton (27:46)

And there's some things.

‍

Bob Sutton (27:47)

We talked about standardization earlier and it's a weird analogy, but we just did a case study of transformation at the Department of Motor Vehicles in the state of California, which it doesn't seem related, but there's this guy, Steve Gordon, who is a tech executive who did it and his team made all these top down changes and rules and an allocated budget for digital transformation and all over the place.

‍

Bob Sutton (28:10)

But then he visited all 180 field offices and collected ideas and encouraged people to come up with solutions that were then spread throughout.

‍

Bob Sutton (28:19)

And I thought that was a great example because he and his team made a lot of authority-based decisions, but then they went sort of bottom-up to collect ideas and to do some local prototyping too.

‍

Rebecca Hinds (28:29)

So Arvind, when we think about bottom-up change, how do you as a leader think about activating bottom-up change and ensuring that everyone within your organization is able to use these technologies and use them effectively so the real impact.

‍

Arvind Jain (28:42)

In the enterprise happens when everybody in the company is adopting the new technology, AI has to come to people for it to be fully sort of become part of their day-to-day work.

‍

Rebecca Hinds (28:56)

And is that equally the case for AI agents versus traditional AI assistants?

‍

Rebecca Hinds (29:02)

Does it become more important?

‍

Arvind Jain (29:04)

It’s better to embed agents also in your workflows?

‍

Arvind Jain (29:08)

So as opposed to thinking that I do some type of work and I'm going to actually invoke an agent to do should become natural.

‍

Arvind Jain (29:17)

I'm in the system where I do that work and the agent automatically comes in and adds value to me.

‍

Bob Sutton (29:23)

So that's a really interesting tool to me because it's a standardized tool that enables hyperlocalization all at once.

‍

Bob Sutton (29:30)

And it's also woven into the nature of the work that they're doing.

‍

Bob Sutton (29:34)

It's not a separate thing.

‍

Bob Sutton (29:35)

So those to me are two hallmarks of things that actually stick as opposed to this thing you do off to the side, that isn't really my work.

‍

Rebecca Hinds (29:42)

A lot of these tensions fundamentally come down to leadership, right?

‍

Rebecca Hinds (29:46)

How do you lead your organization through this type of change?

‍

Rebecca Hinds (29:49)

Arvind, I've heard you say that AI has changed how you lead, how you manage.

‍

Rebecca Hinds (29:53)

Talk to us about that.

‍

Arvind Jain (29:55)

Yeah, well, so first, just for me personally, you know, I've, you know, I have a, you know, one thing I learned is that, you know, I'm a difficult person to work with.

‍

Arvind Jain (30:05)

Like, people have tried to tell me that like for the first 20 years of my career.

‍

Arvind Jain (30:08)

I never knew that.

‍

Arvind Jain (30:09)

And so it's a new learning.

‍

Arvind Jain (30:11)

And one of the reasons why I'm difficult is because I just keep asking questions.

‍

Arvind Jain (30:15)

I don't stop and like, you know, and these questions actually have a cost.

‍

Arvind Jain (30:19)

Like, you know, when I work with my direct team and ask them a question, they take it seriously.

‍

Bob Sutton (30:23)

Like, I was not really like thinking that seriously when I actually post the question.

‍

Arvind Jain (30:27)

But then sometimes, you know, people do two weeks of work to actually answer that question, which I was, you know, was just a curiosity that I had.

‍

Arvind Jain (30:32)

So for me, but it's important I need to understand how the business is doing.

‍

Arvind Jain (30:36)

So I still have those questions.

‍

Arvind Jain (30:37)

And what AI is now helping me with is it becomes, you know, that first stop for any question that I have and I go and I can ask, you know, arbitrarily complex questions.

‍

Arvind Jain (30:50)

Here, for example, how's this part of our business doing right now?

‍

Arvind Jain (30:54)

What are the key risks that we have for our engineering team?

‍

Arvind Jain (30:58)

Or I'll go and ask AI that, hey, go do some research on the web and tell us some new things that we should be building.

‍

Arvind Jain (31:05)

So these are all complex questions that sometimes I would ask our Chief Product Officer or our Chief Technology officer and I'll consume a lot of time for them.

‍

Arvind Jain (31:13)

But now I'm self serving myself, so that's one big change for me.

‍

Arvind Jain (31:17)

Like, you know, when you can bring all the knowledge in the world, all the knowledge inside your company and like, you know, with that amazing reasoning and thinking capabilities that these models have, it's sort of like, you know, I have these infinite set of like, you know, amazing teammates, you know, that are very knowledgeable and I would rather first go to them because, you know, it doesn't take that valuable time away from the rest of my team.

‍

Arvind Jain (31:37)

So that's one big change I made, which is making AI the first colleague that I'm going to actually get my work done for.

‍

Rebecca Hinds (31:46)

Bob, you've studied something called the amplification effect or the magnification effect in terms of how a leader's actions do have these massive outsides.

‍

Bob Sutton (31:56)

Oh yeah.

‍

Bob Sutton (31:57)

So whether or not it's interesting, I'll make a comment a second.

‍

Bob Sutton (32:02)

But whether or not leaders realize that there's all sorts of evidence that when you're in a position of power, people are watching you much more closely than you are watching them.

‍

Bob Sutton (32:10)

And this happens with baboons too.

‍

Bob Sutton (32:13)

It turns out that, what do you call it, the lower-level creatures watch the alpha male a lot more than the alpha male watches them.

‍

Bob Sutton (32:19)

And some of it's number and some of it's adaptive.

‍

Bob Sutton (32:23)

But the thing from your story that makes for a great leader that I like is I would go back to self-awareness, and it's back to the baboon troupe.

‍

Bob Sutton (32:32)

It's understanding what about them when they're watching you, what bothers them and what you need to amplify.

‍

Bob Sutton (32:40)

And that's exactly the kind of self-awareness that I would think is important.

‍

Bob Sutton (32:45)

It's interesting the degree to which you said I ask so many questions, I have to be careful because it starts rattling people.

‍

Bob Sutton (32:51)

But you still need answers to those questions to do your job.

‍

Bob Sutton (32:54)

The other thing that I think is even I think about what's really important in Age of AI, that's always been important.

‍

Bob Sutton (33:01)

But even more important is because there's just so much noise.

‍

Bob Sutton (33:08)

This is one of the main things that almost comes out of our report.

‍

Bob Sutton (33:10)

There's inconsistency in noise, like what's our strategy?

‍

Bob Sutton (33:13)

What do we do?

‍

Bob Sutton (33:15)

Is AI good or bad?

‍

Bob Sutton (33:19)

So to me you talk about the importance of having strategic intent, knowing what your strategy is.

‍

Bob Sutton (33:24)

And the other thing is, especially as organizations get larger and more complex, there's all this evidence that leaders have to say fewer things and have to say them over and over again.

‍

Bob Sutton (33:35)

It's even more important now so people know what to focus on and what to ignore because it's one of the noisiest times I've ever seen in business and I've been watching businesses for a long time and so I'm a.

‍

Bob Sutton (33:48)

So when you're pretty sure you have the right strategy and the right work practices, say the same thing over and over again no matter how bored you get as a senior leader, I think that's really important right now.

‍

Rebecca Hinds (33:59)

I think that's a great note to end on.

‍

Rebecca Hinds (34:01)

Arvind Bob, thank you so much.

‍

Rebecca Hinds (34:03)

What a great conversation.

‍

Rebecca Hinds (34:04)

If you thought so too, subscribe so you never miss an episode.

‍

Rebecca Hinds (34:08)

Thanks for spending your time with us and we'll see you next time on Intelligence Real and Imagined.