AI Madness: Nicolai Tangen's Journey to 100% AI Adoption
Nicolai Tangen, the CEO of the Norwegian Sovereign Wealth Fund, dives deep into his radical approach to integrating AI in asset management, claiming to be the “Madman in Chief” of AI. He shares how he achieved 100% AI usage across the fund, focusing on relentless internal communication and education to drive acceptance and implementation. We explore the transformative impact of AI on productivity, with Tangen citing impressive gains in efficiency that would keep the fund competitive in a rapidly evolving landscape. He also addresses common resistance points, emphasizing the importance of understanding and overcoming compliance concerns to harness AI’s full potential. Whether you’re leading an organization or just trying to keep up with the AI wave, Tangen’s insights are invaluable for navigating this new frontier.
Takeaways:
- Nicolai Tangen, CEO of the Norwegian Sovereign Wealth Fund, emphasizes the need for relentless advocacy for AI within organizations.
- He believes that companies must embrace AI fully or risk falling behind their competition, as compliance can kill innovation.
- Incorporating AI successfully requires a mix of technical knowledge and a willingness to adapt, which can lead to a culture shift in organizations.
- Tangen highlights the importance of continuous upskilling for employees, indicating that mandatory training sessions have been effective in boosting AI usage.
Companies mentioned in this episode:
- Norwegian Sovereign Wealth Fund
- ChatGPT
- Claude
- Cursor
- Morgan Stanley
- Goldman Sachs
- JP Morgan
- Anthropic
- OpenAI
Transcript
Hi, this is Adam Davidson, host of the FeedForward podcast.
Speaker A:This one is really fun.
Speaker A:We're talking to Nikolai Tangen.
Speaker A:He's the CEO of the Norwegian Sovereign Wealth Fund.
Speaker A:It is the single largest pool of money ever assembled.
Speaker A:And Nikolai, perhaps more than any executive we've spoken with, has gone AI mad.
Speaker A:In fact, he specifically calls himself the Madman in Chief.
Speaker A:He's achieved pretty much 100% AI use, 50% advanced AI use.
Speaker A:And on today's podcast, we go through how he did it.
Speaker A:Hi, Nikolai.
Speaker A:So great to have you on.
Speaker A:Hey, so, Nikolai, what is your exact title?
Speaker A:How do we refer to you?
Speaker B:Well, I am the CEO of the Norwegian Sovereign Wealth Fund.
Speaker A:I have called it, when I was a business reporter, the single largest pool of money ever collected.
Speaker A:Is that too dramatic or is that accurate?
Speaker B:I think it's pretty accurate, yeah.
Speaker A:And an amazing story.
Speaker A:I was in the Middle East.
Speaker A:I was a Middle east correspondent.
Speaker A:I reported on the oil industry.
Speaker A:And you see, in country after country, getting oil and gas leads to bad governance.
Speaker A:It becomes a cancer in a country.
Speaker A:And there's always an asterisk that Norway is different because of the Norwegian Sovereign Wealth Fund, that rather than having oil and gas as this kind of corrupt, free for all, it becomes this incredible resource for the people of Norway.
Speaker A:So it's exciting to talk to you, and now you're in charge of maintaining that.
Speaker A: I feel like right now, end of: Speaker A:And sometimes it's two feet on the brake and no feet on the gas, depending on the company.
Speaker B:Well, we go both of the gas.
Speaker A:You got both on the gas.
Speaker A:And so I want to start.
Speaker A: nderstanding is it was around: Speaker A:Can you walk me through what that conversion process was?
Speaker B:Yeah, you know what?
Speaker B:It really kicked off with the.
Speaker B:With the launch of ChatGPT.
Speaker B:And I have a.
Speaker B:Well, we have a podcast here in the fund.
Speaker B:And I got Sam Altman on the podcast, and, well, and I had Bill Gates on the podcast, and I have Jensen Huang on the podcast and Daria Mode on the podcast.
Speaker B:And he was just like, oh, my, there is some stuff happening here, and it's really accelerating.
Speaker B:And the more work I did, the more I saw that it was just, like, totally transformational.
Speaker B:So then I just started to bang on about it internally here.
Speaker B:And I think for this to really be a success, you need.
Speaker B:You need to have a total maniac at the top who just, every time he gets the microphone, just talks about it, and he talks about it all the time, and he never stops.
Speaker B:And it's totally relentless, you know, so that's important.
Speaker A:And that's you, just to be clear.
Speaker B:That's me.
Speaker A:You are the maniac.
Speaker B:Yeah, I'm totally.
Speaker B:I'm totally.
Speaker B:And.
Speaker B:And then you need.
Speaker B:I think everybody you hire, they need to be very, very, you know, kind of computer literate and, and, you know, technical.
Speaker B:Not necessarily.
Speaker B:They don't need to code necessarily, but they need to.
Speaker B:To be sharp.
Speaker B:They need to know ideally on physics and maths and, and that kind of thing just to understand these things.
Speaker B:Then you need to have.
Speaker B:Then you need to get it down through the organization.
Speaker B:And so we have ambassadors.
Speaker B:We have.
Speaker B:We are an organization of 700 people.
Speaker B:We've got 50ambassadors who are being trained every week by, you know, some of these large tech companies.
Speaker B:And then you need events and we have like tech week, hack, hackathons, coffee talks.
Speaker B:So you basically just bombard the organization with information and pressure.
Speaker B:And on top of that, you need to educate upskill.
Speaker B:And the thing is that that cannot be voluntary because the people who need it the most, they just don't particularly want to do it.
Speaker B:So for us, it was.
Speaker B:You had to do it.
Speaker B:We made kind of a seven program.
Speaker B:Upskilling.
Speaker B:We did have seven upskilling sessions, starting from the beginning, really, and then going all the way to the top.
Speaker B:And now we have started on the second stage of that where we have two different avenues, one for less technical people and one for the super super pros.
Speaker A:I want to go a little slower and dig into some areas.
Speaker A:So one thing we hear a lot is resistance comes from.
Speaker A:Often comes from three areas.
Speaker A:The legal team is worried about risk exposure, certainly with customer facing that might not.
Speaker A:That might apply less in your case.
Speaker A:You know, will the AI, will we lose our IP protection?
Speaker A:Will we accidentally promise something to our customers, or will we accidentally be racist or something terrible in a chatbot?
Speaker A:Chief financial officers seem obsessed with is there a return on investment?
Speaker A:And in my view, that's often premature for a transformative new technology.
Speaker A:You might see what they call a J curve, where there's a diminishment of productivity, possibly, although there's evidence that's not even true.
Speaker A:But I don't know when you really want to calculate roi.
Speaker A:Sometimes that's premature.
Speaker A:And then very often it is really freaked out.
Speaker A:They're freaked out about all the security vulnerabilities.
Speaker A:I also wonder if they're not sometimes freaked out just politically, that it's.
Speaker A:Power in an organization often comes from control, effectively from saying no a lot.
Speaker A:And if you're distributing this unclear, imprecise capability, computer based capability throughout the organization, it sort of goes against the nature of it.
Speaker A:So what, what breaks, what resistance did you face?
Speaker A:How do you overcome that?
Speaker B:Yeah, I love, I love your three categories.
Speaker B:Yeah.
Speaker B:Yeah.
Speaker B:I mean, the first one is legal.
Speaker B:You know, there are.
Speaker B:I've never seen a situ.
Speaker B:So, okay, if you apply this, which we do, we think we improved productivity by 20% per year.
Speaker B:You know, if you have, if you have a company here which is applying it, and let's say you gain a 50% productivity gain over two to three years, competition, who doesn't apply, they will, they will never catch up.
Speaker B:You never catch up.
Speaker B:These type of things.
Speaker B:And so basically, and if you lose 50% productivity versus competition, you are dead.
Speaker B:You know, you're bust, you're bankrupt, you cannot recover.
Speaker B:And so it's the first time I've seen compliance being able to kill a business.
Speaker B:They can kill it.
Speaker B:And the reason why they do it is because most often they don't understand.
Speaker B:And you basically kill what you don't understand.
Speaker B:You know, you just pull, pull the thing.
Speaker B:And the problem is nobody can overrule compliance.
Speaker B:The CEO cannot overrule compliance.
Speaker B:Nobody can.
Speaker B:So if you don't play ball with compliance and you don't have an understanding compliance department, you.
Speaker B:You are finished.
Speaker B:The second thing on cfo, yes, these things are costly, but the IT cost is instead of wages, it's instead of people.
Speaker B:You know, it's interesting, we had at the board the other day, we said, Nikolai, your IT cost per person is like $140,000.
Speaker B:I said, yeah, but it's not an IT cost per person.
Speaker B:This is instead of people.
Speaker B:This is.
Speaker B:If we didn't have it, we would have had to have twice as many people.
Speaker B:Right.
Speaker B:Much more expensive.
Speaker B:So I think if you kill it off for financial reasons, that's.
Speaker B:That's just totally moronic.
Speaker B:And the last thing about the IT department, well, hey, you know the cool thing, this basically takes the IT department from having been something boring with cobweb on, and you basically lift it up to be the new heroes in your, in your company.
Speaker B:And that's what we've done.
Speaker B:You know, last three years, IT department, they were in a cupboard.
Speaker B:Now they are, you know, they are the stars.
Speaker B:And so that's the one thing.
Speaker B:The second thing is that is turning kind of the age profile upside down because, you know, in the old days you got some grads, some graduates, and it was just like, oh, my God, this is going to take so long.
Speaker B:Time to train them.
Speaker B:They're not going to be in production within three years.
Speaker B:Now, bang, they're in production within 45 minutes, you know, and they're super sharp and you can apply them to projects.
Speaker B:This is very cool.
Speaker A:I don't have evidence for this.
Speaker A:This is my gut sense, based on lots of conversations, is that the IT folks who are most resistant also know the least.
Speaker A:So they can't be the rock stars of AI if they don't really know that much about AI.
Speaker A:And so it sounds like you have an IT department that was able to adjust very quickly, either had that knowledge or acquired that knowledge very quickly.
Speaker B:Yeah, they were very, they were very fast, right.
Speaker B:And we, I dragged them along to, in the beginning, to all kinds of things and they.
Speaker B:But they were just super fast in doing this.
Speaker B:And then what we also did was we established kind of an AI, an AI group.
Speaker B:And so we now have, you know, a group of people that's the only thing they do is just coordinate projects across the firm so that we can guess, get the best practice across.
Speaker B:We also have a really cool dashboard.
Speaker B:Everything we do here, we log what we do.
Speaker B:We.
Speaker B:I can go in here and I can look at where we, where our IT resources are spending the times, what type of projects, in what department, what proportion of our IT development force are we.
Speaker B:Are we applying, where.
Speaker B:But there is another one here called AI project.
Speaker B:So I can look at what, where all the efforts on the AI programming side is going in.
Speaker B:So that's really cool too.
Speaker A:All right, I want to jump to something else you said about 15, 20% increase in productivity.
Speaker A:Some of our members have said is measuring productivity has forced them to realize they don't actually know how to measure productivity of knowledge workers, that it's.
Speaker A:I remember one member talking about the.
Speaker A:They have factories and they have knowledge workers.
Speaker A:And at the factories, it's fairly easy to say how many widgets are we producing every minute, every hour, whatever, how much input, how much output.
Speaker A:But with knowledge work, what is a more productive finance person?
Speaker A:What is a more productive IT person?
Speaker A:The metrics aren't clear.
Speaker A:So how do you feel that confidence in those ROI numbers?
Speaker A:Because you're all knowledge work, right?
Speaker A:You're not secretly manufacturing.
Speaker B:Well, privacy law.
Speaker B:Privacy law is very strict in Europe.
Speaker B:So you can't measure anything.
Speaker B:It's not like you can't measure, you know, the number of times I'm touching my, you know, my keyboard.
Speaker B:Yeah.
Speaker B:So we did.
Speaker B:Just did a really rough one.
Speaker B:We.
Speaker B:We asked people, what has it done?
Speaker B:What has this done to your productivity gain last year?
Speaker B:And we saw the biggest increase where we expected to see it.
Speaker B:I on the programming side and the development side, and then less in some other areas.
Speaker B:And on average last year it was 14.7%.
Speaker B:So I pretty.
Speaker B:Pretty close to 15.
Speaker B:And the funny thing was that I asked Sam Altman, you know, on the podcast, so, Sam, how much productivity do you think we can gain here?
Speaker B:And then I just took a number out of the hat, which is like, do you think we could get to 10%?
Speaker B:And then he just asked, well, how many people in your firm is doing this?
Speaker B:How many are doing that?
Speaker B:And then he came up with 20.
Speaker B:So we ended at 15.
Speaker A:Wow.
Speaker A:Sometimes I wonder, like, productivity obviously matters, but another thing that AI, in my own personal work, sometimes I'm less productive because I spend more time researching lots of things, learning a much wider range of things, spending time wrestling with difficult ideas.
Speaker A:And I could imagine a situation where an investment analyst, say, who in the past read a series of reports and came to some type of conclusion and reported it up the chain.
Speaker A:One version would be AI, allows them to synthesize reports much more quickly and they're more productive.
Speaker A:Another version is AI, allows them to see far more detail, synthesize way more detail.
Speaker A:So maybe they go from spending, I don't know, a week to spending three weeks on the same report, but the report is much higher quality.
Speaker A:The investment decisions are much better informed.
Speaker A:I wonder how you think about that.
Speaker A:Productivity versus growth and new opportunities.
Speaker B:Well, I'm not sure one.
Speaker B:I mean, one doesn't exclude the other.
Speaker B:I think you do both.
Speaker B:Right.
Speaker B:We have developed a tool internally.
Speaker B:It's just meeting preparations, where you just stuff in all your previous meeting notes a structure for how you normally would like to conduct a meeting.
Speaker B:And then in addition, it just kind of drills down and does deep research on the topic.
Speaker B:And so instead of spending three hours, you just get it ready.
Speaker B:So that saves you time to potentially, you know, do other things and read up a bit wider and broader on other things.
Speaker B:So I think it just enables you to go wider and broader.
Speaker A:Yeah, absolutely.
Speaker A:For me, for sure, there are things that used to take me five days that now take me an hour, but there are also things that used to take me five days that now Take me a month because I'm going much broader and it's back and forth.
Speaker A:You said another thing that I wanted to dig into about the need for technical knowledge, I want to dig into that a bit.
Speaker A:I am finding that the more I'm not a coder, but I create a lot of software these days and the more I begin to understand at least the basic structure of how Python works or how a development, you know, the key things to think about, how testing works, how front end and back end interact, it definitely makes me stronger, I understand more.
Speaker A:But one of the sales pitches of AI is it actually unlocks non technical people to be able to, to engage the world in a new way that you can.
Speaker A:You know one of the experts at FeedForward, Matt Bean, has done this amazing research into what he talks about as shadow learners.
Speaker A:That poorly educated person on the factory floor who maybe had an insight but didn't have a way of communicating that up through the organization.
Speaker A:But now AI allows someone who maybe doesn't even speak English or Norwegian or whatever it might be, but can communicate far more clearly.
Speaker A:So I just wanted, I wanted to dub, just understand better what you mean about that technical knowledge because I'm wondering if that is necessary in all cases.
Speaker A:There might be people without technical knowledge who can really flourish in an AI world.
Speaker B:No, you for sure can, but we are just seeing that people who have a technical background are grasping this.
Speaker B:They seem to be grasping them a bit faster and they seem to be able to kind of make them interact with the rest of the systems we have and perhaps they just understand processes and how to split a processes in a better way.
Speaker B:I mean we have 100% usage, you know, of.
Speaker B:We use Claude across the organization, but to really develop new things.
Speaker B:It seems like people with a technical background has, they seem like they're doing.
Speaker A:Better here and I think that systems thinking, that is something I am really on fire about right now is that you can take, you know, it's forced me to realize I have a bunch of systems I use.
Speaker A:I was a business journalist for decades and I never thought of them as systems.
Speaker A:I just do my job and now when I break it down and I can create little agents that represent different stages and then I can think cleverly about, hopefully cleverly, maybe stupidly, but I do my best of when do I need my brain in there?
Speaker A:When can I just have a process run?
Speaker A:But that process thinking, that systems thinking seems so crucial.
Speaker A:And for sure computer programmers are just trained in that way, but so are other people.
Speaker A:Scientists are Trained in that way.
Speaker A:Some academics are trained in that way.
Speaker A:And that makes a lot of sense to me.
Speaker A:I want to get to that hundred percent number, and I want to dig in more deeply on what the maniac in charge needs to do.
Speaker A:But before we get there, I will say some of our members are not you.
Speaker A:They're not the leader of the organization.
Speaker A:They're a level or two below that leader.
Speaker A:And they're looking at a C suite that's a little unsure, and maybe a board that's unsure.
Speaker A:And that is thinking, all right, we're hearing a lot about AI, but we heard a lot about Bitcoin, and we heard a lot about other things.
Speaker A:Is this real?
Speaker A:Are the risks worth it?
Speaker A:And maybe if we go slow, maybe we have to just go, you know, step by step.
Speaker A:And maybe you're the wrong person to ask because you're the boss, but how would you manage up if you had to?
Speaker A:How would you get that boss to get religion on this?
Speaker B:You tell you you just need to make it blindingly clear the seriousness of this situation, you know, because if you don't act and if you don't get started, you're just going to lag behind and die.
Speaker B:And then when you propose this, if you get a no, then you just don't take no as a no.
Speaker B:I mean, no.
Speaker B:When people say no to me, it doesn't register in my head.
Speaker B:I don't know where it goes, but it never goes into my head.
Speaker B:And I just think that when.
Speaker B:If people say no to me, it's just because they haven't quite understood my question or what I'm saying.
Speaker B:So, you know, in this Diplomat series, this Netflix series, Diplomat, you know, they say no, it's just a rest stop on the way to yes.
Speaker B:And that's kind of my.
Speaker B:That's kind of my philosophy.
Speaker B:So you just need to totally make it clear to them what's going on, you know?
Speaker A:Yeah, I just finished that series last night.
Speaker A:I really enjoy it.
Speaker A:Yeah, walk me through.
Speaker A:So.
Speaker A:So if you're sitting with a CEO, you're at Davos or something, and.
Speaker A:And the CEO of some big Fortune 100 company in America is sitting next to you and saying, yeah, we have an AI initiative, and we have some people looking into it.
Speaker A:Of course, we're.
Speaker A:You know, we have our compliance people making sure we don't take any unnecessary risks, and we're moving deliberately and slowly.
Speaker A:We don't want to be front of the pack.
Speaker A:We're happy to be a middle or a little behind the middle.
Speaker A:What.
Speaker A:What would you say to them?
Speaker B:I would you say, Adam, sorry to say, no offense, but you need to wake up because competition is stealing a march and you would never catch up with them.
Speaker A:Yeah.
Speaker A:Speaking to the converted.
Speaker B:I know, I know.
Speaker A:Yeah.
Speaker A:I, I spoke to my daughter's school the other day and the teacher said, we're going to wait three years.
Speaker A:And I was like, what are you talking about?
Speaker A:If you said we're going to wait three weeks, I'd be confused.
Speaker A:Years.
Speaker A:You're going to just let these kids grow up with AI?
Speaker A:You know that means a freshman's going to be graduating high school with no guidance from you.
Speaker A:We already saw what happened.
Speaker A:That's the other thing I like to say to these folks is everyone's using AI at your company.
Speaker A:It's just.
Speaker A:Do you know that they're using AI and if you think compliance matters, then they're using it in a fully uncompliant way.
Speaker A:I do love what you said about compliance can kill.
Speaker A:And I, I like the idea that on what.
Speaker A:I mean, there are very real compliance risks and companies do have to pay attention, but the number one risk should be the risk of inaction.
Speaker A:Right.
Speaker A:You said it's existential.
Speaker B:Absolutely.
Speaker A:It seems like one thing that is crystal clear, I'm guessing at the Norwegian sovereign wealth fund, is your career success here is going to at least partially be determined by how you use AI.
Speaker A:That there's no question that this should be used.
Speaker A:That.
Speaker A:That culture change also feels like something.
Speaker A:I hear a lot about that.
Speaker A:It's a question I ask at your company.
Speaker A:Is it embarrassing to say you used AI to help you with a report, or is it embarrassing to say you did not use, like, I.
Speaker B:Well, I would say, I would say if you didn't use it, you'd be, you'd be totally stupid.
Speaker B:And who do you think is going to be promoted?
Speaker B:The people who produce twice as much as the other person, or vice versa.
Speaker A:Yeah.
Speaker B:So, but, you know, but you can't produce just a report.
Speaker B:And you can't, you can't just produce a bad report.
Speaker B:You have to have a human in the loop here and make sure that it's outstanding.
Speaker B:But, you know, it should be well written.
Speaker B:The cool thing with these, with this is you can just kind of, you can just turn the dial in terms of what kind of complexity do you want to have to come out?
Speaker B:And so, you know, if your report is for people who need 15 years of education, then that's where you turn the dialogue.
Speaker A:Yeah.
Speaker B:And you just increase complexity so that's a really fantastic thing here.
Speaker A:Yeah.
Speaker B:So easy.
Speaker B:That's so easy.
Speaker A:Is it explicit?
Speaker A:Like, how do you.
Speaker A:I mean, it sounds like it's in a million different ways, but getting that message out that you, the way you're using AI need is going to be a part of your career advancement, your opportunities here, et cetera.
Speaker A:Is that explicitly stated?
Speaker A:Is it just obvious because the CEO is just talking about it all the time?
Speaker B:No, we tell you that, we tell you, say all the time.
Speaker B:Isn't if you, you know, you, you don't want to make it a threatening thing.
Speaker B:So you, you're not going to say it's going to take your job, but basically people, people who use it are going to take your job.
Speaker B:Right?
Speaker A:Yeah.
Speaker B:And so you, I think you really need to use a combination of carrot and sticker.
Speaker B:I think it needs to be mandatory to, to upskill and to take, and to take the classes you need.
Speaker B:At the same time, you don't want to lose out and so you want to, you know, you want to be in the forefront.
Speaker A:Can you talk a bit more about the carrots?
Speaker A:What are rewards?
Speaker A:Do you do ex.
Speaker A:You know, Ethan Malik often will say something like if, if the CEO just had a sack of $10,000 every morning and just dumped it on the desk of whoever did the most with AI in the previous 24 hours, it would save the company a lot of money over time.
Speaker A:It would be the best investment they could make.
Speaker A:I'm guessing you don't do that.
Speaker A:But are there explicit ways do you celebrate successes?
Speaker B:Yeah, yeah.
Speaker B:You know, I would say incentives is that if you do a good job, you get promoted faster than other people.
Speaker B:Right.
Speaker B:So that's a huge incentive to use it.
Speaker B:But we also highlight good user cases and successes when we got town halls and those kind of things and you know, stories on Jammer, which is our internal kind of Internet equivalent.
Speaker B:So, you know, you get a higher profile in the firm.
Speaker A:Yeah.
Speaker B:We put you out to give lectures to universities and, and other organizations.
Speaker B:So it's a, it's a huge incentive to be good at this.
Speaker A:That's great.
Speaker A:Can you walk me through some of the projects?
Speaker A:Just what are some of the ways you're using AI that are particularly effective or just interesting?
Speaker B:Yeah.
Speaker B:We are the most cost efficient asset manager in the world.
Speaker B:I think it's partly because we've got a lot of assets.
Speaker B:Right.
Speaker B:But we also are extremely efficient in how we run it.
Speaker A:I mean, 700 is not a lot of people to run the biggest pool of money.
Speaker A:In the world?
Speaker B:No, no.
Speaker B:And we run it.
Speaker B:We run the money at a quarter of the cost of equivalent organizations.
Speaker B:And we use it for, I mean, gee, for just so many things.
Speaker B:Everything from how we deploy capital around month end.
Speaker B:We typically have inflows at the month end and how you prioritize the trading.
Speaker B:Also, how do you, how you net your buying and selling?
Speaker B:We got, you know, very many mandates and some people, you know, perhaps you want to buy Tesla on a Monday and sell it on a Friday.
Speaker B:So these models basically predict a lot of this and enables you to net the buying and the selling against each other.
Speaker B:So you save a lot of money.
Speaker B:We use it to monitor our investments.
Speaker B:We have live.
Speaker B:We own 9,000 companies around the world.
Speaker B:You know, we own one and a half percent of all the equities worldwide.
Speaker B:And we monitor these 9,000 companies live on, you know, the news feed for these companies is being automatically graded in terms of severity of news items.
Speaker B:And we do this in 16 languages, you know, live.
Speaker B:And you know, this would have been taking just an unbelievable amount of resources in the past.
Speaker B:If we want to, if we meet with a company and want to check the quality of the board, as you just press board quality, it basically looks at the board members, it looks at what kind of relevant experience do all the board members have.
Speaker B:And bang, you have a score for that.
Speaker B:We vote on nearly 100,000 proposals every year at AGMs.
Speaker B:And you know, we just put them in with our expectation documents and get suggested suggestions for just how we should vote compared to how we have voted in the past and how we typically vote for these type of things.
Speaker B:I mean, it's just endless amount.
Speaker A:Are there asset categories that just the due diligence would have made it impossible?
Speaker A:Like maybe emerging markets or really small firms or some asset category that the due diligence outweighed any potential benefit in the past, but now you have access to new types of investment.
Speaker B:I'm not sure I have a good examples of just that, but it for sure facilitates and makes all types of due diligence better.
Speaker B:You know, I've got this podcast, I've got a business podcast where I interview all these leaders and let's say now I want to have a podcast with a bank.
Speaker B:And so I put in my previous podcast with, you know, the head of Goldman Sachs and Morgan Stanley and all that kind of stuff.
Speaker B:It knows my kind of, you know, my preference for what it should look like.
Speaker B:It digs up all the extra information and then at least I have an outline and it makes sure.
Speaker B:That I don't forget anything, which is quite important.
Speaker B:Right.
Speaker B:You don't want to have any obvious questions which you haven't covered.
Speaker B:And then, of course, you slim it down and you kind of work with it and so on.
Speaker B:But it's a very, very good starting point.
Speaker A:Yeah, that's right.
Speaker A:Actually, I'm embarrassed to say I don't do that.
Speaker A:And I, of all people, should do that and.
Speaker B:Oh, you should do that.
Speaker B:You should.
Speaker A:Yeah, yeah.
Speaker B:You should not have spent a lot of time preparing for this thing.
Speaker B:You know, it's just like stuff in all your previous podcasts and just put in my name and just say, hey.
Speaker A:Go for it, Go for it.
Speaker B:You would get a pretty good outline, you know, for sure.
Speaker A:Yeah, I. I am genuinely embarrassed.
Speaker A:I don't do that.
Speaker A:Like, that's exact.
Speaker A:I have a lot of tools like that.
Speaker A:I just haven't applied it to the podcast yet.
Speaker A:I just downloaded this new tool that is a meeting agent that called Cluli, and it just tells you what to ask.
Speaker A:I mean, but that's just for business meetings.
Speaker A:It has these little buttons, you know, what should I follow up on?
Speaker A:You know, was that guy.
Speaker A:Was that an evasive answer?
Speaker A:How should I press them?
Speaker A:You know, that sort of thing.
Speaker A:It even has a feature which is not appropriate for me, but where if you're in a technical interview, it can answer the technical question.
Speaker A:So if you're in a job interview and you ask me some technical question, I can act like I knew.
Speaker A:Knew the answer.
Speaker B:What is it called?
Speaker B:Cluly.
Speaker A:Cluly.
Speaker A:Yeah.
Speaker A:C L U E L Y. Cluly.
Speaker A:Yeah.
Speaker A:So we don't have to do any work ever again.
Speaker A:And then we could get 11 labs to do the interview and in our voice and, you know, we could just go sleep in.
Speaker A:You do have this great platform of the podcast and that.
Speaker A:I mean, to me, that's interesting for a bunch of reasons, not least of which is I've been working in podcasting for almost 20 years, but it seems like a part of your role as sort of chief internal AI evangelist is really understanding this thing and putting yourself in conversations with people who can expand your thinking.
Speaker A:I'm just curious about that part of it.
Speaker A:I mean, I think it's a space where every few weeks there's something pretty transformative.
Speaker A:Every few months, there's something massively transformative that I think should rewrite what we imagine these things can do and how we can use them.
Speaker A:And I would guess that not every CEO is built to adjust that quickly and stay that curious?
Speaker A:How is that?
Speaker A:I mean, you strike me as someone who likes living, as Ethan calls it, on the jagged frontier.
Speaker A:But tell me about that.
Speaker A:I mean, the way you think about AI today has to be very different from a year ago.
Speaker B:Sure, I do think a lot of CEOs are, you know, one of the features of many CEOs are very curious and they're very keen on knowledge and they are in a situation where they speak to a lot of important people who are very, you know, very much on the forefront of knowledge.
Speaker B:And we are lucky to do that too.
Speaker B:So, you know, we speak to many of the important CEOs, but I would also say the relationship we have with, let's say anthropics is very important because you get a glimpse into what the next launch is going to look like so you can get compliance to, you know, prepare and, and so on.
Speaker B:So when it's launched, you can slot it in straight away and that, and that's important.
Speaker B:And then we are seeing, you know, from other people, seeing user cases and we, I mean, I would say we, we steal with pride.
Speaker A:There's a nimbleness, I guess, is what I would say.
Speaker A:I mean, I.
Speaker A:Most CEOs I know I are, yes, they're, they're curious, but there can be a rigidity that this is, that they see their job as, here's our strategy, here's our strategy.
Speaker A:And it's hard to embrace a radically different approach.
Speaker B:Yeah, it's interesting, you know, the more, you know, I went with some good friends up in the mountains the other week, and so these are the cleverest people I know.
Speaker B:And we were sitting there and we were predicting what was going to happen next year and we are tape recording it, right?
Speaker B:And so then we played the tape recording from last year.
Speaker B:Oh my God.
Speaker B:I mean, these are clever people.
Speaker B:We were like 80% wrong.
Speaker B:Yeah, we had, this was just before the election in the U.S. we, we didn't expect, you know, the outcome.
Speaker B:We didn't, you know, we didn't know of the tariffs or the relationship between, you know, the superpowers and all these kind of things that should be going on.
Speaker B:You know, what's been going on with Fed and the Bureau of Statistics.
Speaker B:I mean, gee, who could, you know, who could have predicted this?
Speaker B:It's not easy, right?
Speaker B:And perhaps you did, but not many people.
Speaker B:So the thing is that you forecasting is just useless these days because there's just no way.
Speaker B:And the thing, and the funny thing is, if you told me, hey, this is exactly what's going to happen, I would have thought.
Speaker B:Okay.
Speaker B:Stock market's down 25%.
Speaker B:Well they were up 15% on nearly 20.
Speaker B:Yeah.
Speaker B:So you can't predict.
Speaker B:So the whole thing is about being fast and agile and that's what you have to do.
Speaker B:And so that's why you need to put these tools in place so you can just move a lot faster.
Speaker A:That makes a great deal of sense.
Speaker A:I want to dig in on Anthropic has a pretty aggressive plan to build financial analysis tools within the models.
Speaker A:Within their models.
Speaker A:I assume OpenAI and Google do as well.
Speaker A:I don't know.
Speaker A:And do you have that thought that more and more of your business gets absorbed by these core foundational models?
Speaker B:I suspect.
Speaker B:I mean we are not big users of them yet.
Speaker B:We haven't really gone there to a high degree.
Speaker B:I suspect they will be.
Speaker B:I suspect, you know, to be build your own models and spreadsheets and so on will be pretty outdated very soon.
Speaker B:I think making investment returns will be about other things.
Speaker B:I think it's too early to say.
Speaker B:I'm not sure how good these models are just yet, but I'm sure they will be fantastic at some stage.
Speaker A:Yeah, I go back.
Speaker B:The thing is it will take away all the repetitive and boring tasks so you can spend time on a higher level thinking.
Speaker A:I have to assume that if I went to the Norwegian sovereign wealth fund or JP Morgan or whatever and really studied how do people actually spend their time all day, every day that we'd be able to find a whole bunch of tasks that will be fairly commoditized by the models.
Speaker A:But that at least for the foreseeable future there will be a frontier between the best in class model and the best in class investment mind.
Speaker A:Right.
Speaker A:You know, it might be that investors, human investors get far more specialized in what they do and that by bundle of things they do gets broken up and there's a whole bunch of things they don't do anymore.
Speaker A:But they really look for that keen insight.
Speaker A:It'll be very interesting.
Speaker A:I go back and forth personally.
Speaker A:On the one hand I sometimes think anthropic and OpenAI and Google are a little too arrogant in their imagination that they can take over entire industries by hiring a few finance people say to build finance tools within their model that you certainly employ people who are probably better than those people they hired and your people are certainly better at addressing your particular financial needs.
Speaker A:But then the models do seem to acquire a generalizable skill set as they get bigger and as the scaling laws multiply.
Speaker A:So it's probably one of the most essential questions of our time.
Speaker A:Right.
Speaker A:Where does human value come from?
Speaker A:What's enduring in that?
Speaker A:How long does that last?
Speaker A:How does that adjust?
Speaker A:I wanted to get a little bit more micro in how it actually works.
Speaker A:So you said everybody has access to Claude and that's the primary tool that you use.
Speaker B:Well, we have Claude for everybody and we got 100% uptake.
Speaker B:We also have Cursor which we use and where we have roughly 50% uptake and then we have Copilot but we use that less.
Speaker B:But you know, to have 50% of the firm using cursoris.
Speaker A:Yeah, that actually shocked me.
Speaker B:Yeah, yeah.
Speaker B:Because I just saw, I saw Morgan Stanley was making a big deal out of.
Speaker B:They, they say, well, 20% of our staff are kind of developers.
Speaker B:What?
Speaker B:50%, 50% of our stuff of developers and that's, you know, a financial institution that's afford manager.
Speaker B:It's pretty incredible actually.
Speaker B:So that is how hard we pushed it, you know.
Speaker A:Yeah.
Speaker A:And just for people who don't know, Cursor is what's called an ide.
Speaker A:It's.
Speaker A:I mean until a few months ago the only people who would ever use it are software engineers.
Speaker A:It's very intimidating if you're not a software engineer and you first see it.
Speaker A:You can have a cloud code or any AI chat built in.
Speaker A:Cursor has their own, that's very good.
Speaker A:But it's a lot of just raw text code and it can be.
Speaker A:I mean I'm in Cursor all day, every day.
Speaker A:I love Cursor but it took me a while.
Speaker A:It's pretty scary.
Speaker A:And when I show it to non technical people, there's a huge barrier to overcome.
Speaker A:But it sounds like.
Speaker A:So it's not just developers who are.
Speaker B:Using Cursor, it's a lot more people.
Speaker B:I guess they're kind of turning into developers.
Speaker B:But yeah.
Speaker A:Do you use it?
Speaker B:No, I don't use it personally.
Speaker A:You know this new one, Cursor two that just came out, I would recommend it because it now has these easy to build agents.
Speaker A:You could just think through a simple process and in a half hour build a Nikolai machine that at least does some component of your job.
Speaker A:But I'm very impressed with that.
Speaker A:And is that in the training, in that seven levels of AI training, is Cursor part of it or did that spread virally?
Speaker B:Well, no, that's part of, that's part of AI 2 to 0.
Speaker A:So yeah, so that's what you're launching now or that's the well, people have.
Speaker B:Started to, you know, a lot of people have already totally in it.
Speaker A:So to me that is a huge step.
Speaker A:The going from AI as chat to AI as tool builder opens up a huge range of new capabilities.
Speaker B:So, you know, I sat next to this developer on the plane the other day and I try always to speak to people next to me on the plane.
Speaker B:I mean, you don't always want to start the conversation too early because sometimes you sit next to some really, really weird people.
Speaker B:So I said, thanks to this developer.
Speaker B:He was like, he was so excited.
Speaker B:He's like, yeah, you know what?
Speaker B:I've, you know, I've written four apps today and on average I've just, I've just typed less than half a percent of the code and the rest is just automatically generated by the likes of Cursor.
Speaker B:Right.
Speaker B:It's pretty cool.
Speaker A:Yeah, it's pretty cool.
Speaker A:I'm studying Tibetan and I was online looking for flashcard programs.
Speaker A:So to help me with.
Speaker A:I didn't find any good ones.
Speaker A:So this morning I just had Cursor create me of my own customized flashcard app.
Speaker A:That's pretty good.
Speaker B:I mean, how long did it take?
Speaker A:Half an hour.
Speaker A:I mean, and then another 20 minutes to make it perfect for my purposes.
Speaker A:And it's pretty nice.
Speaker A:If you saw it, you'd think it was.
Speaker A:It's actually, I would say it looks better than some of the commercial tools that I saw.
Speaker A:I'm doing that every day, all day long.
Speaker A:I mean, I just, you know, I can't even count the number of programs I've created.
Speaker A:It's.
Speaker A:I love it and I don't know code.
Speaker A:It's thrilling.
Speaker A:So when you say 100% though some number of that is mostly chat.
Speaker A:Using it for probably kind of what I would call entry level AI and then it goes up from there.
Speaker A:I'm guessing that there's sort of a.
Speaker B:Well, it depends.
Speaker B:When I walk around the office all the time, right, Just to see what people are up to and have a chat and get information about all kinds of things.
Speaker B:It's pretty, pretty complicated.
Speaker B:I would say it's relatively advanced.
Speaker B:A lot of the things they're doing.
Speaker A:Yeah, that's great.
Speaker A:I spoke to a rather huge company recently and their internal survey found 1% had used more than just chat, which I found shocking.
Speaker A:Yeah.
Speaker A:How do you do the training?
Speaker A:Is that in house?
Speaker A:Did you hire a firm?
Speaker A:Is it AI based training?
Speaker B:No, we developed it together with an outside.
Speaker B:What kind of consultant?
Speaker B:You know, a very, very good speaker.
Speaker B:And so we, so we Developed it ourselves and.
Speaker B:But with some kind of, some help from the outside, just in terms of the presentation and so on.
Speaker B:We had seven modules, half an hour each.
Speaker B:You kind of get a QR code at the end so that we can see that you've done it.
Speaker B:And you know what, people hate to be forced to do stuff.
Speaker B:They hate it.
Speaker B:But even if you do something which is 100 correct, 10 would always hate you anyway.
Speaker B:So you just have to live with that kind of, with that kind of noise.
Speaker A:Yeah.
Speaker A:Has there been natural.
Speaker B:And actually, and actually it's quite fun because, you know, what we can measure is the usage of, you know, of Claude.
Speaker B:And we could see after each session the usage was going up.
Speaker B:Incredible.
Speaker B:You see this staircase, right?
Speaker A:So they maybe some grumble about it, but they're still using it.
Speaker A:Yeah, totally, totally.
Speaker A:Have you lost.
Speaker A:Have people left because they just don't want to be on this bandwagon?
Speaker B:No, we haven't.
Speaker B:But you know, there is, there is a bit of a stretch here.
Speaker B:I mean, there is, there is dispersion, right.
Speaker B:And you have people who are not very keen on it, who resent it and they will be less happy about life generally.
Speaker B:So that's why you really have to get them on board.
Speaker A:You know, I mean, we often talk about wanting to identify those early adopters as a natural beachhead to build capability, to build an internal lab, etc.
Speaker A:But yes, eventually you really do need that broad use.
Speaker A:And it's those last folks who have insights that AI could help them express with us.
Speaker B:25% are pretty.
Speaker B:Are moving quite early and then you got another 50%.
Speaker B:Those are the people we are really working on with the next level of AI education.
Speaker B:The last 25% are.
Speaker B:I don't really know what's going, what they're doing, but they will move as well because you just don't want to fall too far behind.
Speaker A:Right.
Speaker A:My last question that I'm guessing is on the minds of some of our members.
Speaker A:You have a lot more capital than any of them, but you have a much smaller staff than any of them.
Speaker A:You know, these are companies with hundreds of thousands.
Speaker A:Some of them have millions of employees.
Speaker A:How would you adjust if you were suddenly named, you know, CEO of.
Speaker B:You.
Speaker A:Know, some massive company with hundreds of thousands or millions of employees?
Speaker A:Does the same rule book apply?
Speaker B:Totally.
Speaker B:I would just, I would just apply exactly what we've done.
Speaker A:Maniac CEO who talks about very, very.
Speaker B:Conscious about it in audio recruiting.
Speaker B:You know, force people up the, the skill levels and give them all the tools they want.
Speaker B:And then just events and internal communication.
Speaker A:Yeah, that is very scalable.
Speaker A:And by the way, and 30 minute training sessions, those are bite sized.
Speaker A:No matter how much you hate it, it's over pretty quickly.
Speaker A:And if you actually are learning something, that makes a lot of sense, great.
Speaker A:All right.
Speaker A:Nikolai, it's been such a pleasure talking with you.
Speaker A:It's exciting to hear about everything you've been doing.
Speaker A:I'm also excited that you'll be able to meet a bunch of our members.
Speaker A:I believe it's next week that we're doing a live chat with you and the membership.
Speaker B:Very good.
Speaker B:Look forward to stay in touch with you guys.
Speaker B:All the best of luck.
Speaker A:Thank you, Nikolai.
Speaker A:Great.
Speaker B:Thank you.
Speaker B:Thanks.
Speaker A:I hope you got a lot out of that conversation with Nikolai Tangen.
Speaker A:You may also have attended the live event we did with him.
Speaker A:He's a great guy.
Speaker A:We plan to have him back for other events.
Speaker A:Please let us know what you want to hear about on the podcast.
Speaker A:You can DM me or email me or Jessica or Maddie.
Speaker A:Can't wait to see you in the Discord.
Speaker A:And at our next event.
