How Startups Can Crush Enterprise AI Sales | E2050
58:36 Share

A serb should be looking at like what what's different and it's like what part of what's differently is we have reason as utility. Think of every role in the economy as a bundle of tasks. In those tasks, some of them fall to A I and some of them for to the humans.

So I do think what serves should be looking at us. Like how can I change the way like if you look at workflow, the way that it's been built over the last ten years, it's been built with humans at the center ah and that's the actual workflows. So if you were to zoom all the way out and say, okay, well, i'm going to have a workflow be shared, some of this gonna be through an algorithm, some of this gonna through a human, what would that allow me to do?

This weekend startups is brought you by lincoln ads to redeem a one hundred dollar lincoln ad credit and launch your first campaign. Go to lincoln dog com. slash.

This week startups then desk. The best customer experiences are built with sandeep k. Qualifying startups can join their starter program and get zen dex products for free for six months. Visit sendest stock com slash twice today to get started and beehive power er news letters with A I tools refer programs and add network features all in one platform get thirty days free and twenty percent off your first three months at behaved dotcom sas twice.

Everybody, welcome back to this weekend. startups. This is alex, and I ve have a special show for you today. There's a theme going around the world of startups, which is A I height, A I fundraising, A I applications, A I agents, A I chat ts, A I job loss, A I this, A I that.

What i've always been very curious about is behind the headlines, how is enterprise generated A I spend going? And thankfully for me, even your capital, a firm that i've known for a long time, min law stopped a very long and detailed report doing. And due precisely that, how is generated A I doing in the enterprise? So I decided to have a couple the authors of the report come on the show to answer my questions.

And then in a stroke of fortuity luck, minda a is perhaps best known in the A I game for being one of the backs ers of anthropic. And this morning, the company announced to raise four billion dollars more from amazon's. And now we have the vcs from the report trapped under on thun so we can hit them and all sorts of questions as we would like. But just welcome them into the show. First of all, we have job reform.

Job, how are you? And fantastic links.

perhaps me on the show. absolutely. And what I like the most about you, apart from the fact year part of the report is that you were um at yahoo back from two thousand three to two thousand and nine.

right? And I was I was early yahoo prior to be a product of my entire career, right beno about a year ago I was a cheap product of her at last year. I was there for six years.

And then before that, I was an early vice president of product at living din when I was a tiny company, pri po, and helped grow that up in to a ten thousand employees. I was there for seven years. Yeah, I love building things. Uh, start companies, sold the company, hope, take a company public. That's a bit about me.

Yeah yeah. But the cool thing is that you worked yahoo where i've also worked.

And it's like of all the things, alex, that we can point out. It's like I don't know if yahoo is the one, but I yeah no yahoo in the early days and you will remember this was just super fun. Um IT was like IT was really the google of the day as the internet was emerging back and in the late ninety. So it's it's always a tRicky one because I I think you know Younger folks and a that are listening really know yahoo as the the amazing company the that we were part of in the early days of be uh like a an amazing ing institute like uh, harvard was of a sudden, you know not harvard IT was at a different tier. So change quite a bit of the years.

It's like if harvard became and how the state .

was purchased by private equity .

and I was to go with, yeah that's a deep cut. I'm in red ireland, which is why all northeast jokes work. But we also have dirk show with us. Dirk ky, you, my friend, I was prepping for the show. And the thing about your background that I love the most is that you were president of the harvard.

That's sorry, I started in journalism a lot of overnight, I think with venture capital but ah trying to yeah it's an interesting start and um a lot of admire a lot of work that and .

excited be on that. Um also dark on your minda page, you are tagged in on the anthropic inside of things. Now I know it's mat Murphy who's the lead partner over atman look for the anthropic relationship.

But today the company announced that IT raised four billion more from amazon, bringing up to a total of eight billion dollars and I I guess the but I wanted know is why why is IT still so expensive for these AI companies? Because that's a lot more money, not that far afterwards. And to me, sitting here I am, apart from the numbers as over IT feels a little terrifying. But I was hope of you to swatch my oh my .

god at that number. Yeah I mean, I think anthropic uh have a special relationship with amazon. I feel like they have a very close partnership um and this is just a further end of that.

But the other side of the coin is that anthropic is one of the foundation models, right? This is when we made our first and investment back in twenty twenty three, the thesis was that this was going to be one of the, you know, companies that will matter the most in the A I. revolution.

And we seen that plays out. That's why we uh, have been getting closer, closer to the company. And you know the announcement today probably just a furtherance of that and deepening of the relationship with the amazon.

So jaf, you guys took part in the series c and then if I were call correctly, a LED of the series d .

is that right? Yeah, we through sp, we were lead on five hundred million of the billion dollar rays that happened .

on the d side. And I take IT now feeling pretty smart, given how A I has gone since the deal was put together.

Yeah well, we we're excited to share with you some of the findings that we've we've learned on the L M side. Um as IT contains to the enterprise right obviously two large markets for l lam. We have the consumer based market and then we have the enterprise based market and IT, it's some you know IT starting to shake out that these are different markets and in how competition is approaching an attacking notes are quite different. A very, very happy to see some of the data uh, behind the anthrops brokers on the on the on the enterprise side. I know that i'm slightly .

harassed with the anthropic news when I asked you to come on to talk about the actual interprets A I report, but i'm just current one more question about that.

Does having a company like anthropic in the broader family portfolio, does that really bring a lot of information to the investment team the you guys can then I don't learn from and apply directly to making new investment decisions? Or is that information segregated from the firm? And see.

you guys can't go fishing in that point to where very church stay when IT comes were were anthropic and in both men lower. Um you know we're not in their deep looking at people's use such metrics or anything like that。 But I will say one benefit of the money that comes from that is just being able to see what actually is coming down the the pipeline.

When IT comes to new models, we forced the relationship, a special relationship with the anthropic. Five months ago, we announced our andoo gy fund. That's a hundred billion dollar fund and that's really looking for um who are some of the most pioneering A I founders out there and being part of that uh fun gives gives founders A A number of benefit.

Early access to models, twenty five thousand dollars and uh and property credits, access to some of the deal teams over there and some of the expertise you be part of a network of fellow founders and builders in the A I space. I and then every so often once twice a year, we we run a builders say our first builder day with anthropic was uh the first of november and that was just that was super forms that was at the anthropic offices. We had a bunch of the experts uh from anthropic coming out, meaning with uh companies that were building on the building on the platform.

And yeah, so we were wearable to have a really successful builder day. So I would say, you know, that part of the relationship is really helpful for both of us. So we really charge the the team over there and there.

They're just amazing. I mean, the the the team and the the caliber of the talent been routing anthropic. I just I I admire every day it's it's really quite talent well with .

four billion notes more they can certainly ep caring. And I know one of your predictions in the report, continuance of the A I talent to jump. So we get about in a minute. Let's start with what everyone wants to know, which is the the high level numbers.

So you guys wrote in this two thousand twenty four the state of generated A I in the enterprise that A I spend in searched to thirteen point eight billion this year, up six x from two thousand and twenty three. Now before we get into into categories, Derek, did that match your expectations? Is at a faster piece of growth then anticipated to me big number, big jump. No idea if I was a bullish er berries compared to your projections.

yeah. I mean, I think that when we did this report last year, we thought that this would take a little while to uh, rap up. And that's what you saw in previous technological transnational like this, right? You benchmark to cloud, you'd benchmark to mobile.

And these are all you know trillion dollar kind of, uh, markets. Now that took a little while to yet started. And so I think when one of our predictions from last year was that this would similarly take a little longer to um to ramp up despite all the spending that's going to IT.

Have you built something awesome? Are you ready to bring IT to market? Well, that's great.

But you're probably asking yourself, how do I get to my target customers? Well, here's an easy solution for you. It's called linked to that.

And with linton, you can precisely target the professionals most likely to care about your product. You can use lindon to reach people by job tittle, industry, company location and more stuff that you have on your profile page. Imagine you're looking to build old europe to b audience, maybe you want mid science companies, maybe want large ones.

May be you want emerging small businesses, right? Small and medium size businesses. S, M, B, S. Well, you got got to build those key relationships and you want to drive results.

Will you just do that with linton ads? And you got to gain access to over a billion members of linked in, of which thirteen percent hundred and thirty million of them, our decision makers. That means they can pull the trigger on buying your product service.

And ten million of them, one percent of the lincoln audience are those elite ite sea level executive. You know what i'm talking about? Chief technology, chief strategy officer, chief for revenue officer, chief executive officer, chief financial officer, they're all there on linked in and you're going to get two to five times the return on your outspan.

Seventy nine percent of B2B mar ket. Same linked in delivers the best paid media results for a reason. So here's call to action. Start converting your B2B aud ience int o hig h qua lity lea ds. Today i'm going to give you a hundred old credit to start your next campaign linked in the com flash this week, start to claim credit that's linked in the com flash this week. In startups, s terms and conditions do apply.

I think what we've seen this year actually is the a surprise in terms of half that the matter revenue is actually materializing. And so you look at areas like health care or legal, which are actually traditional lager in terms of technological adoption. And these are actually the vertical that are leading the a the a revolution in many ways on the apple. Um there is you know hundred million dollar plus companies now in both of these verticals. And um I think that that part was surprising is just like how much that apple player has IT has really taken off.

The reason why I actually less shocked by that I IT would be, is that my spouse works in medicine. So i'm actually this, really aware of that world. And interface iena es, but if there are places where you could apply AI to a voluminous out of you written information, legal and healthcare are gotta do the request.

You know, apples on the tree. So even if they are historical larger ds, doesn't that actually mean that they have more accumulated equivalent of like technical debt in their Operations and therefore, they are the best place to deploy? So actually, I can kind of see that, dad, I have less surprised.

I thought that I would. Yeah, absolutely. And the other way that we look at IT to is the comparison size of the 臭豆腐 market verses the size of technology spent there。 So if you think about someone, yeah, I mean this this goes towards you, the silicon valley, a kind of phrase now as services of software.

And I think that IT rings true at a cliched because, uh, of a reason, right? Because these are massive markets, the tech we used to not be able to touch. And now, uh, with A I, they can now also made a lot of that.

You know, one thing that does surprise me, and going back to this come at the different man, is just the of the curve on A I, right? Having lived through a number of these different waves over the years, you know, the internet came and then cloud computing came and then mobile came. They tend to be much slower.

You you lose fact that, uh, the enterprise level IT was really this is the genre movement started two years ago. And I would pick that at january twenty twenty three. IT was january y twenty twenty three.

The ChatGPT came out and said, hey, we got one hundred million bw faster than any other application out there. Yeah, I was a really telling story. I was I was still at at last thing at that point.

I was that SHE product ops. And in january, like that first uh, earnings call, there was like one mention of A I. But if you looked at all of two, there was in a single brunch of generative A I. So I went from the first earnings called by the second earning call. There were eighteen mansions in the transcript.

So like a fun activity is go grab any company right, and pull all of the transports from twenty twenty two to now and load them up into your favorite at alarm and just say, build me a chart of the number of mansions on A I or artificial intelligence. And what you see is it's like IT goes like crickets to, hey, something's happening here to all a sudden that is the conversation. yes.

Um so I would say that is is definitely one of the things that played out, and that's one of the findings that we have in the report. If you look at twenty, twenty three, that would be the year really of the pilot. So the the C E O gets up the try, uh, the earnin's call with the CFO goes over the R N D D group and says the C T O and the C P O like, hey, what are we doing? The gender of A I.

And very, very quickly what happens is teams get hold together. It's like, how are we going to use this new technology? And that's really one of the findings from the survey is the last year was a lot about experimental. This year is really about moving step into production. And that's where we get the the six ax increase in the overall uh spend at the enterprise level coming up to close to fourteen billion uh by by our Marks.

I want to talk about uh basically data sourcing for the stuff because I love this particular chart. But to when I see this, I go, how confident are you guys in these numbers? Because you could easily categorize things in different pockets. I put there some bleed between them and also is a growing industry. So can you just talk me through the how this time has put together and if you into the audio this shows on europe of your changes in A I spend for foundation models, training, deployment data, vertical department .

and horizon. Let me tell with the high level, i'm i'm going to talk through the ho'd actually got number s in. Really there's two basic buckets that you should be focusing on.

One is around the l lam in the infrastructure needed to bring A I into the organization. And then the second pocket, these four are actually three three bars on the right are really talking about the application layer. So we can see is that two thirds of the spending goes on at the enterprise.

Nine point to eh billion is sitting in the l lambs in the infrastructure bucket. And for those they can see by far the largest spend inside of the enterprise is against the foundation models themselves. Six point five billion of that is spent there.

So two thirds is happening at the end infrastructure layer and then we have another uh, third, which is being spent at the application layer. The main thing about the application later, the big story is there is the that's an ax increase from from the prior year. So they are sitting at four point six billion.

Then what alex is, is asking one of the question is asking is like you can categorize your application layers a lot of different ways is um in this case we've broken IT down into vertical departmental and horizontal A I and and certainly we can swap about like wigs in the water bucket. But I would say the higher or story is that applications are coming their alive and well. And then for a variety reasons, we've chosen, and to break IT out .

this way in terms of data sources, we went out and ask six hundred I T decision makers. So these, who are you budget owners that have her view over their organizations generate A I spend. You know this is very lengthy survey, but basically how much are you spending on generate A I what is that relative to your overall spend?

Um and then very specifically, what tools are you spending IT on? Um I think that one of the motivations behind the survey was that there's just a lack of good data out there of like water actually I T decision makers really kind of looking out what are the cases that they have um and words their money going towards. And so this is um you know we spoke to six hundred. We didn't go get to all you know the fortune thousand. And so there's obviously this is direction, our answer um but I think that it's pretty informative to see some the insights of like what folks are actually, you know not just playing around with anymore but actually adobe.

So just the four point billion for the different types of a applications. I should probably pay a bit more attention to the year of your growth in that number then to see here and go. Why is there four point .

six and all four point seven billion actually? Okay, great. What are they doing at that application layer? Or you know in the thing that I would say is fascine about that is that there's a real broad set of gena use cases in the enter Price.

So sometimes you think see things get very concentrated around one department or one use case, but I think IT speaks to the the usefulness of the technology. So when we break down, if on average and enterprise has ten geni uh use cases among st and what which ones are most popular? Yeah and that would be that be an obvious question. So when we look at the code generation was number one, fifty one percent of the folks on the survey I said that they're using A I for code j, the building software for job.

Just to be clear, not fifty one percent of the six hundred who are paying for genres of A I services, but of all the six hundred, fifty one percent companies are pain for code completion tools. yes. okay.

Just want to make sure so this is as big of a number of this a hundred as we can matter. okay. I think that .

keep that's not surprising. We've seen um get get up. Uh copilot is one of the you know fastest growing revenue products out there.

And in the application layer of A I, there are north of three hundred million. And uh in annual revenue, we've seen players like cognition, kodiak all hands come about. I'm after code, jen is support chap pots. Probably not surprising there thirty one percent are saying they're deploying chap pots. We've got products sa deck gon, uh, a sera, which is from I T S M I T server management use case a is a big one there.

Hey, startups, when you're a business, you gotto treat your customers right. Unreasonable hospitality is the standard today, but you can need tools. You going to need a platform to help you do this.

And that platform is the same desk sweet. The sen dest is onna. Give your start up all the tools you need to deliver exceptional customer experiences so you can build stronger relationships without growing your headcount.

That's key, right? Every dollar matters. You ve got a control account, you got a control spend.

So use the tool. That choppy square space, uber and installed all rely on. It's called that. Let's take a look at another customer unity, very famous company. They saved one point three million dollars with the desk, automation and self service, and they saw an eighty three percent increase in their first response time.

These companies loves them that because it's so easy to set up and IT skills with you as you grow the also give you all the metrics to make your reporting easy, keeping you and your business agile and investor ready. And that's because your they are received customer and they've created the zendelwald artus programme just real. We get unlimited access to all the endesa expert insights, all the best practices and entry into the amazing community of founders, all at no cost for the first six months.

That's right. They want to support you zena. Stop, come, slash, twist. Get ready to scale with the best in customer support. With six months freed, nothing to lose.

Then he move to enterprise search retrieval, only eight percent. Um you know these products like a product like clean has been around for a while. But with the emergence of geni, what they have been able to actually accomplish on the first front has been um dramatically Better than baby, what they were able to do in the early years.

Another product and there would be something like sona then we to um you know things like uh meaning summarizing uh you know so many of us are remote workers and we're so used to seeing that A I agent has been added to either polar transcript or summize ation of our conversation yeah um things like firefight in in order and they know the last one I think that round out the top five would be would be copyright writing. And this goes back to comment that you are making out like, you know, elms are like calculators for words there. Like good I had authoring contents, say, you know, you could certainly see a world in which everyone and would be using uh, or having a copywriter editor, you know, grammar checks, just part of whatever the communicating to help them be more precise and conscious. Products like writer type face, where we have an investment copy A I, they tend to be some of the leaders in that space. But though that gives you a feel for like I could get what is going on at the application layer of the enterprise.

So a lot of words is my reading that I think the calculator for words is a great think about this. Code generation is creating characters on the screen, sport top pots, create characters on a screen, you know, meeting symbolization, ation characters on the screen, CoOperating characters on the screen.

The thing that surprised me is that workable automation was so low in this chart to date because if you go back five years, we're thinking about U I path and you know, R P A rebuck process automation. And there was a lot of enthusiasm for generated the I prel EMS that R P A was going to remove a lot of the human dredging from digital work. Then we got much Better tooling.

But now they were, you know, looking back at two dozen twenty four, i'm not seen as much of that show up as I kind of expected. And this is A A long way of saying how much progress have we actually made on agents. I suppose that can take some of the work from us and do IT versus being more assistive in our day to work and duck, I see you blinking.

I be so go home. yeah. I, I, I think that we think about IT in terms away. And so the first ways of G A I apps were what we call brag apps or a retriever augmented generation based um and so usually you know you you have a external l knowledge store and you uh you use IT for things like synthesis is so if is one of our portfolio companies, uh it's illegal copilot.

And what IT doesn't takes um a lot of long gant legal tax and makes a generates reports that generates uh legal grief ing things like that so that the lawyer on the other and doesn't have to go through the Georgey of all that work. And so that's kind of the first generation. And then as you get more advance, we move into agencia architecture, which you know today, our survey found that IT is a minority. But if you ask a year ago, if you look at our previous report, IT didn't exist a year ago and this idea of you had like baby A G I and auto gen, some of these like you know open source projects back then. But I didn't exist as an enterprise idea, something that you know, when I think about traditional R P A, like your ipad, um the idea of applying its enterprise haven't really existed and now IT does so.

So dirk IT sounds like what I was doing was just been impatient and now IT is showing up. And so what I expected to happen is I just had my time lands of mentally compared to the market.

I think the technology is now there. And if you look at things that we're really excited for in twenty twenty five, this is one of the things that we think explode is moving from retrieval based architectures to more agents um and things that can were automate work close across orizondo areas. So like you, thank you.

I ipad, but also vertigo st help care. There's solution like color that are doing uh kind of in just automation and um you know a lot of different uh for domain specific applications. I think you you also say.

and i'm literally right now point up a the OpenAI O N preview blog was because I forgot the term that I need here. Exact time, serious thinking when models take a little more time before they make a decision or return to prompt IT. Is that underpinning the improvement in the technology that is making the agenda a approach more feasible today?

Erp, yeah test time inference. And so I think they I think that there is a couple layers to this, right? Um so you can have stuff like a one which is kind of formalizing a design um uh kind of pattern at the model layer.

And so as a model gets smarter, IT makes fewer areas. I think the problem with the agents that you had traditionally is that it's kind of run in a recursively. So if you think about IT, you know the agent will the agent will think OK in order to accomplish this, like tough.

That requires ten steps. What are those ten stuff? S and then i'll go out and do IT, and then I think I might step one, okay, check what just step to.

And then if you think about err rates there right all ate um that now will established if you have a ninety nine percent a rate or ninety nine percent accuracy rate. And step one, if you haven't that for all ten steps, by the time you get to step ten, you know that array is something that is unacceptable for Price. And so that traditionally been the problem.

Um and so alliance getting smarter is part of the answer. But also when you apply to specific domain, get a scaffold IT. We we like to use the term asian on rails, which is basically you need a hard coat or harness of the domain of, like all the actions that the agent can take. You need to kind of set guard rails on IT with hard, with code in order to point you and get, you know higher levels of the accuracy so that that resist area rate doesn't compound too much.

Do those gardens have to be programmed on a per use case, per industry or per company basis? I'm trying to figure out how hard IT is to hard. Good because to me that could be incredibly complicated or relitigate easy.

I just don't know where IT lands. Yeah I think keep trying to figure that out, right? And I think that obviously, the more specific, particularly use case, the higher accuracy and more robust IT is and but the are treating off really is degrees of freedom which like way at this very and is A G I, you know no guard rails, just a model, just like put IT in a four loop in that runs.

And then on the other end is um what we have today with just computers that our hundred percent are coated application logic is determined by uh, the computer or first of all, some programmer who SAT down and was like, okay, i'm thinking in the shoes of the user, what do I need? And so I think the answer will probably be somewhere in between. We're trying to move towards H.

G. I eventually. But yeah, I think right today, we need a stricter yes.

I think you know there's a here in a doctor there. There is a good example in the software, gentle space OK. So let's take code generation. And we can say, well, how much progress are we're making in code generation.

One way we can look at progress as we can look at the score called the sweep entry, right? And we can look at that over the last what's happened there in the last eleven month. So swe bench for those that don't know, it's um it's testing real world tasks faced by software developers.

And the benchmark is based on things like pull requests and issues from open source github or posit ori. And I think there's something like twenty two hundred tests in there. So if we go back to january this year, about four percent of those tests were completed by the best software agents, uh, system out there.

In march, there is a company called cognition, which got a lot of while attraction. You might remember that by march, I was fourteen percent. Now if you look at at the number one score on three bench belongs to one of our portfolio companies, is an open source, often are company called all hands. And they can solve fifty three percent of those cases.

Hey, founders, let's talk about a super weapon that your start up might be missing. Killer newsletters, news sites aren't just a great way for you to stay relevant. They're also an incredible growth engine for your startup social iphone forms in search traffic.

Those things are losing their edge, which your email list that remains your direct line of communication to your customers. You don't have to pay a middle man. You don't have to get caught up and try to trick the algorithm m into getting your message through.

But in order to have a successful newsletter, you need a world class platform. And that's why we moved the this weekend sort of newsletter to behave. Behav is a newsletter platform that's built to help you grow your community and turn your newsletter into an absolute powerhouse.

It's packed with amazing features like an A I driven post builder. It's also got this really clear rebuilt in referable program. And a top notch ad network if you want to turn on modernization.

Plus, they have really easy to use tools if you want to charge access fees like subscriptions. So here's a quote, action b hive plants started just thirty nine dollars a month, super affordable. But we've got an exclusive deal here here at this week.

And startups, a thirty day free trial, followed by twenty percent off your first three months. That's right. Go to behaved dogs like twist to claim that twenty percent off behaved dogs twist to start growing your news letter today.

So if you kind of look over the eleven and months, right, it's gone from four percent to fifty three percent. So that's that's like, uh, you know thirteen percent improvement from the beginning of of the year is dramatic.

And I guess you know, jeff, how confident are you personally not speaking for any other company is that, that rate of progress can be kept up for the next twelve of eighty months? Because if you run, the numbers are long and up. Eventually we get damn close to a hundred percent.

Yeah I look, I think it's gonna be uh in very IT will I will depend on what department and what use case that you're moving against will move from fifty three percent to a hundred percent next year on the software agents. You know I do think IT a you know the stuff fairly and optimized right now. So I think even what the models that we have, let alone new advancements in the the underlying foundation model, I think there's tons that can still be done from an accuracy perspective.

okay. So thinking about the categories of enterprise G A, I spend that we started with discussions about improvements to agenda I and the approach, therefore, and also improvements to existing leading categories. This all sounds very bullish on the technology side. My question is how much of the revenue they were talking about is spread out amongst the startups because know job you said that you know, six and a half billion other third one point eight is foundation models spend OpenAI and throw a couple of names. When we get to the four point six million from the outside, the number of companies that are nimble at that spend is enormous.

So is, is there enough enterprise generated A I spend for apps today to support number of starts of that are going after those dollars? Or are we like is there half as much spin as we need? I just don't know. This is too much bread, not of butter. I guess you .

know it's a merchant. If we look at the the software budgets at the enterprise for A I and twenty twenty three, they were going toward something called an innovation budget a lot of time. That's like, hey, we just need to be investing in here. We don't have a permanent budget to pull from.

Um so I I do what I do you see happen is you're gna pull from the permanent budgets in time as the R Y uh improves the you know another point that I would make is that historically a lot of the budgets that were being spent by both the business unit and I T. Were wrong tooling. And back to this comment the Derek had made or earlier, it's um you know now we have services of software. So it's not just the tooling budget but it's the human capital budget that we start to move into as well. So once can I explain .

that to people because I think it's a good point, but I want done on IT. So essentially, if if you can reduce headcount and replace that with software spend, you can often get quite a lot of being for your buck. And so budgets for software for A I and I bring place some human activity can be tally, uh, rich because humans are costly in in health insurance and travel types and office space.

And I know I don't think it's just about human reduction, right? Like I do believe that there's their companies are also faced with like a lack of having enough people like I don't have enough software engineer. So what allows me do you know, exchange and continue to grow that apartment, but I can do that in a different way.

So it's not just about reducing its also about expansion. But but that is part of the story. So I I think there are plenty of what to chop in there when in the report itself, you'll see the A I spend by department um and you know you'll see departments like the legal department is historically like not spending money on technology.

They're much more of a later doctor and here they are being um more of an early adopter. So I think the the budgets s are very promising and um I he'll continue to grow as the R Y I continues to prove itself out. So from an R R Y perspective, last year was less clear what the R R Y was this year as we move from for pilots in the production there.

There's greater clarity um but still some question Marks. Sorry, like it's not all figured out the number of folks that on the survey, I want to say IT was something like, uh, you know a third of the survey respondents are saying, well, we're still figuring out our exact implementation on a strategy. But Derek.

picking up on that point, when we think about A I generation that I spent in the enterprise, we essentially asking what what is the for that today? And I was just talking about, uh, on the show of the other day, but how uber, when he was very Young people were companied against the taxi market, turned out B, S. Because the mark of larger IT IT seems that if we can unlock spin from departments that didn't have technology budget to begin with, like legal, for example, the time for A I generate A I enterprises huge, but also IT makes the overall time for software itself larger. And I don't think I actually thought that was going to be the case, but that does seem very bullish if I .

understanding this car. Yeah, I think that generate V I as far as adoption today is more expensive than replacement. And so you haven't like innovation budget last year.

It's actually really interesting. We ask where are you pulling uh uh budget from and of the permanent budget types. Um a lot of IT or over half of the permanent budget for general vi is coming from new budget.

So IT is not you know I am either you know replacing spend for my system of record or some other software. IT is I am creating a new line item for this. And I think that you're seeing that across departments, which is IT is most expensive as well as across a lot of different places that used to not spend a lot on technology. And we're seeing from the start of side, which is very often I spend a lot of our time, you know, we're seeing companies pop up all over the map, whether it's verticals, uh, we entrant health care, legal, financial services or departments, sales and marketing, uh, data science, uh, human resources like it's it's really all over the APP, which is why we're quite excited about you know.

general thing, I have A G OK. Well, I would love to just vamp about A. I like three hours.

I want nw start. So you guys had a great chart that called selection criteria for generated A I tools. And to my surprise, uh, the highest line I am wasn't cost. Instead, IT was rally.

And so IT seems that when people are approaching A I software, what they want is not to spend as little money as possible, but to have the biggest bang as possible. And so i'm curious, what should startups take away from this particular charge job as they approach the market? That way.

they can see the most success.

No one is ever said that before. There's no accelerator that says build things people want.

Yeah you know you will it's interesting in just sticking with like what would be good advice on the start up front. I do find that you know you want to look at what can you do now that you couldn't do yesterday. And that's really a thing of focus on.

I I remember what I was out linked in one of things I did was I started the mobile group there a lot of my peers for saying, hey, look, we got two thousand pages on the website let's like jam in into this mobile device and my point was like, no, that's exactly what we shouldn't do, like the mobile device has something is unique and its special ah. And therefore, what we should be doing is focusing on highlighting those use cases so that you know, in time, I believe that lincoln would become a mobile company and indeed, IT IT didn't wind up coming to become a mobile company. And the product that we built was quite different.

There was only seventeen screens compared to the the two thousand that existed on on the website. So I I think in this moment, you know you have to look at a serb should be looking at like what what's different and it's like what part of what's differently is we have reason as utility. So you know if I look at um there's a there's a guide, Daniel rock is pressure you panny.

Um he had a number conversations with them and he was killing me. Like think of every role in the economy as a bundle of tasks. And those tasks, some of them fall to A I and some of them fall to to humans.

So I I do think what serves to be looking at us, like how can I change the way like if you look at workflow, the way that it's been built over the last ten years, it's been built with humans at the center, and that's the actual workflows. So if you were to zoom all the way out and say, okay, well, i'm gna have a workflow of the going to be shared. Some of us gonna a be throughout algorithm, and some of us can be throughout human.

What would that allow me to do? Because a lot of these workplace that we see at the enterprise level are very complicated and very tool rich. You know if you look at a an Albert sales team, they have like wolf, different wools and and off of sales force or hub spot.

If you look at software development, there's over twelve of different tools that are hungry and off for the software development flows. So you know you have an opportunity to reign what that workflow should be. So you're you're either are gonna enter in and say, hey, i'm just gonna to a iffy the existing tork flow or i'm gna think about IT at a more first principles level and think about what could be possible given what we know now. And and that would probably be my number one. Uh, my number one suggestion has really get back to the first principles of IT.

okay. So I wanted pick up on that because you guys wrote a last income and dominated the enterprise market with boat on strategies that layer gene I capability onto existing products and job IT sounds much more like you're saying, look, don't do that. Ripped the page out.

Start link. And build from the ground up so dark, I think we call this an air force approach, just a tivo, an software um what fraction of software that exists today needs to be ripped out and started over again because people joke? I don't know what salesforce does we all kind of do but not really and you know not to make fun of alassio but gerais hell so um i'm curious here you know how much that I will not serve.

I have filed tickets and I you owe me lunch for the painting for me through don't want people will the people who they compare olia house. So it's fine. I'm just here as there like a how much of software needs to be rethought in this way? Yes.

I don't know that I have a very clean answer for you know x percent of software needs to be completely thought. But across the board, I think that one of the things that realized this year versus last year is that it's actually not bad, easy to build A I networks. You know a lot of people thought last year like, oh, if i'm a self force, i'm adobe, I can just tack on something A I on top of my existing system of record. And that will .

be my first data already. Because our system of record, you have the back in the data to make your own two models.

the data, the distribution that trust with customers all about. And that is why a lot of people thought, like, is A I really a net new category? Or is IT just a feature on top of existing software? And I think what we realized is actually IT is its own independent category.

And why is really exciting for us? Because that gives the advantage to startups rather than, you know, salesforce coming out with agent force and everybody just being like this is the greatest thing, honor. And so everybody will adopt IT.

There is opportunity for startups because people, enterprises, we talk to try out features like agent force and real wait. This is not what was promised. And so there's opportunities through domestic c workflows, you know a AI native approach to actually make that promise work.

But if we think about the companies that exists today that might struggle to move from the kind of legacy software offerings that make them all their money to AI future, that means that there is trillions of dollar ars in the market cap out there for startups that are being born now with the blinch of paper and the modern tools were discussing to go after. So in a sense, I kind of like think we should short the nice deck and the double. Our investment in dc now will be the best way to get kind of both sides of the back.

You know it's different um domain by domain, right? Like it's it's hard to predict and that's what makes interesting fun like a dobe, for example, they started on the company and then when the club volume came, they actually took a tremendous really stay. If you look at their quarterly revenue, when they decided to go from on crime to subscription, they were able to convert that.

And you know, you could. Adobe is one of the companies that I personally admire a lot. Will I get for them or against them in this gene revolution? I I was a really interesting company. Firefly is a great product, but not all companies will do that, right? And so when you look at the uh starts, but the startups tackling IT and what they offer, where they different from the income as well as who is the in company in the solution or they will position to move with the forces against some um you really have to take a domain by domain.

And IT is like it's a very visual thing when you see a company that gets called up and left behind in that right, you can look at a company like chek. Eighty five percent of the mark cap is disappeared with A I stack overflow site that I use to use quite a bit like their traffic has have as people go directly into the and alone to get their their coating guidance.

Yeah, I wonder how many those we're going to have by this time next year because check made money off helping people cheat on homework and they're going to get back to me about that. So people, please don't email me again. That's when people using check before we all know IT. Now you will use OpenAIr c heat. It's great.

I wish I when I was in middle freshes chemist, did not because I didn't, but I onder how many other companies are kind of on that list? And if that will be a good parameter for how fast AI first companies can kind of up root legacy companies that are I I guess, cloud and sas are now legacy and kind of they almost feel outmoded, like you can remember in salesforce were invented as you like. Oh my god, this is the future. Weird that I was here.

Oh my god, is at the past. You is interesting because I I feel like these are all things. They were building blocks. They were needed to get us to where we are today. We actually needed the internet to get the world's information digits.

We needed to have cloud computing in order to unless vast, vast amounts of of G, P, U and C, P, U. To do training and inference and things like that. So IT IT feels like IT was more of a lining age, building up to the moment that we are today.

Yeah, but how much credit do we give yahoo for inventing the internet portal? Not today and we don't don't even think .

about IT yeah well or z rocks .

in the guy, right?

I mean, it's like you you know you um my company is built on the shoulders of giants or A I is built on the shoulders of giants in a lot of things they needed to happen to get us to the state that we're in today.

Yeah, I think there's building blocks and there's valley capture and I think that a lot of the new value to be captured will be by starts and new companies um but obviously, building on the shouter giants so bad .

we need up here because I have a question about this. I know we're talking about how startups can disrupt in comps, but there are some in combat A I companies. And i'm thinking about very clearly OpenAI on anthropic being two of the largest players in this kind of neo space.

And OpenAI recently put out A A search product, which is Frankly pretty good. I use IT on a regular basis as a testing tool. And I also think that perplexity will see some of its moment and cut because there is now a competing product from a larger, Better finance company. And so when you guys are talking to start up founders, how do you help them navigate building something that we would get stomp on by a foundation model company, releasing something that can be considered to be competing?

I think one of the things to think about is what is on there, what is on the near term roadside, right? And for us, a company like OpenAI and anthropic, their ultimate goal is to achieve A G I for anthropic A G I safely. Um and what are the things that are going to be, you know, the things that tack next.

And so anthropic has already mentioned we won't pursue you sit quest such as image generation or things like that because IT is not necessarily the thing that will bring them to A G I. Better reasoning on other hand, or using tools like web browser, computer use on computer use, those are right. And so a lot of the, you know, when you are in A I APP today, there are two things that you really need to get right.

One is, can I make the base technology work and a lot of the things that actually up to date um or just how do I make the elem reliable? How do I connected into enterprise systems um how do I give IT a tool such as a web browser? How do I know when I was talking about the agents before, how do I make sure that the reasoning is reliable? But you can think about these as data scan falling things that um you know actually are needed today to help uh act as crutches for the L M.

To actually work in enterprise application. Those, I think will won by one, fall away over time as anthropic makes advancements in the intelligence of the base on. But what won't change and topic won't get to because quite Frankly, is not important for A G I is how do I apply this to my, you know health care domain specific workflow if I need to, you know, do like clinical documentation integrity, if I need to do R C M on the back and revenue cycle management and topics is not really interested in that. And so that's an an opportunity .

for outlier startles. okay. But here's my question because all that tracks of medic, I I agree with you whole hardly.

But as we can closure to agi and as we actually maybe you reach IT in the near, near future, doesn't that obviate a lot of the work that's being done to make A I apply to specific categories or market verticals because as the AI brain get smarter, IT probably needs less help to do more. And so I wonder if you will see a solution of the power of going vertical n ai. As we get closer or reach N G.

I, I think the the thing that maybe that discount a little bit or takes for granted a little bit is the training needed to actually make the solution work. You know, if you let's say that you have a super intelligent P H D level human, right?

If you have to if you take that person um and apply them to, let's say, the health sample of like how do you do R C M and work with payers to make sure that your claims get paid, you still have to teach them how to do IT. You still have to you know there's a learning curve as they used to the work close there. And you know obviously, if you take that person verses somebody um you know who may be less intelligent, there is a shorter learning curve there, but they're still a learning curve there. And how you actually uh get that intelligence and applied to the specific worker on the other end, that is the area where application of our companies .

can add value. okay. But you're making A A, A, A, A jump that I wouldn't make, which is you're saying that if you took A P H D level A G I, my hope is that by the time you reach A G, I were no longer using post graduate benchMarks to determine intelligence or expertise. Were so far above that, that those those analogies don't hold and then we won't have to have so much vertical zone guard rails built and so forth because in theory, this should be able to be a bit like .

a magic box and hope this is a question to if there's like, well, when i'll go to rule them all collection lions, lions think my phone. Like the power of my phone is actually sitting in the ecosystem is that there's an APP for that. So there's millions and millions of the apps that make that product more productive.

So you know one version in the world is like I got to now go and IT does everything another version might be there. I have an orchestration algo that knows when to ask other algos like for their expertise in a given area science. So part of that is still unknown um as we move into the future yeah .

I think we we joke about how everything is bungling and unbundling. But I kind of wonder if when we get to agi, if we're going to have ags orchestrating s in the bundle or kind of a singular .

brain mother evolve bundles.

right yeah kind of like if you ve got you know, neth lets and E S P N the same thing, can you imagine that would be crazy?

That would be not.

I know. So some ize, you just kind of taking all this in one in one bucket. Enterprise generation A I spend grew very quickly and perhaps even faster than expected. We are seeing increase spend on foundation models but also on the application side.

And you guys in the middle perspective is that the upstarts are going to knock off more income ants and probably we're going to see more agents C, A, I progress next year. It's going to be very exciting. Is there anything also that the highest level in this report that I have not brought, I want to make sure we get all, all the keyboard meat out, if that make sense of.

you know, there there is one point that I found really fast thing that we didn't plumping to two march, is that when I went into the enterprise and we said, what? What is and this is that the infrastructure. Later we said, what? What is the l that using what they came back and said is that they are leveraging muli models. So I thought that they would try to like get behind a single model um where is for hedging or for performance or cost reasons there they are typically on averaging three different models, which I thought was really fascinating.

And then the other thing is, um you know just the shift in enterprise around what is who is using why from a foundation moral perspective OpenAI clearly had the first move or advantage in the enterprise but always saw on the survey is that they moved from fifty percent down to thirty four percent so lost sixteen, sixteen points here a year. You can see that anthropic doubling coming up. A close source models are clearly, you know, the majority of what's being used. Um meter was black metro, uh, you know, off a little bit, you know.

poor europe. Look that look at their one, A I genti like fourth in line. We got, maybe everyone should spend ten percent of their A I spend on on misal just to help france because, you know, new administration, ukraine. Okay, just me. Right job.

An an word. Investors over here.

I love how uncomfortable he got right there. No, don't exactly.

It's stand clear. Stand down. This is where like that P.

R. Training comes in play. It's like me, is alex beating us? Is this gonna beat the one quote that comes out after all this? Like wonderful, great conversation, you know, alex and team, we're gonna pick on that one time .

that we blame jeff red firm just endorsed all trust administration policies, kids. But I I am I think this very net optimistic and that's a good place to end a year. I think a year of so much change, IT does all feel to be very exciting, like we are really making progress.

And you know, the computer used stuff, tropic, not to give you A A portfolio company extra ops, but like when that dropped, I was as excited about that as I was about ChatGPT, the first time I used to, you know and that that's a great place to be going into a new year of a lot of investment in progress. It's all very encouraging. No, just before you guys go, I do have one lightning round ish question.

So sorry for this, but I I can help myself. So direct start with you. What percent of your new network is in crypto or bitcoin?

I am actually, outside of venture, a very unsophisticated una.

Uh, so a little bit of the etf sounds .

like all index ones. You know that on the U. S. Economy.

bro, you should see my family portfolio exactly the same, and now point one percent by going via just over the same. Q, curious how well crypto explosion you are going into? Twenty five zero zero.

Yeah, this supposed .

be the free bee, not the, not the does take out the quote the very end and make them to the headline, I mean, jesus.

yeah no. I you know I I focus on things where I think I have like Better understanding A A competitive advantage is similar to Derek. It's like passive on the economy and then I am wait over index on a on on venture investing through a brief different funds in anger investing.

So it's like, yeah, if you if you look at that, it's like why you doing that? Because I feel like I have IT like at all. I've not.

I built off for my entire al life that's like that's the thing I know super well. IT has no and it's like not a reflection at all on cypher or my beliefs. There is mostly just about what I am in like I have so much interest in in in start of plan. I appreciate .

the honesty there because I feel like there is always A B pressure whenever crypto does crypto things to appear sophisticated about IT for eighteen months until IT goes away for two years again. Yeah but IT has been quite loud in the the the crypt of fans have been making notes. But uh, anyway, guys, thank you so much and we will come back and do this again next. The two thousand eighty five general A N enterprise of work, i'm sure by them will have even big. Remember to a dig through, but before you go drop yard for handles and they will say, goodyer.

yeah I um I used by linden handle. Good, good. A link in man, that's what I am posting. That's my activity. That's the thing I do to I take with the very enough.

Derek a, what is your preferred social media?

Hand lip, I am, I am on both on twitter. I'm at dirty sha.

all right, very much. And I just back next week for a couple of new shows and then thanksgiving. 我 see them .

by this out。 Super fun.