324: 2025 predictions with Dave Kellogg: The Future of AI, SaaS, and Business
43:49 Share

We're going to move more towards answers, not links, which has been the holy grail for decades. If you want to show up well in LLM results, have your brand mentioned alongside the right words and phrases in authoritative media. Keyword, authoritative media. Well, what's that? Big name publications. How do you get in those? Answer.

The last thing, the last chance for differentiation is branding. Therefore, we need to do branding. And I've heard a lot of people say that. I think it's categorically wrong. Good morning, good afternoon, or good evening, depending on where you're listening. Welcome to AI and the Future of Work, episode 324. I'm your host, Dan Turchin, CEO of PeopleRain, the AI platform for IT and HR employee service.

Our community is growing. We get asked all the time, how can you meet other listeners? We recently launched a weekly newsletter. It contains some additional fun facts and comments from the guests that don't always make it into the weekly show. I encourage you to subscribe. The link will be in the show notes.

If you like what we do, of course, please tell a friend and give us a like and a rating on Apple Podcasts, Spotify, or wherever you listen. If you leave a comment, I may share it in an upcoming episode like this one from Bertram in San Francisco, who is a food tech VC and listens while making dinner.

Bertram's favorite episode is that excellent discussion with former Cisco distinguished engineer and holder of more than 600 patents, JP Vasseur. Great friend. I love that episode about the future of AI for network engineering. We learn from AI thought leaders weekly on the show. Of course, the added bonus, you get one AI fun fact.

Today's fun fact, which is quite germane to today's conversation, comes from Ben Dixon from VentureBeat, who writes about four bold AI predictions for 2025. In it, he predicts the plummeting cost of inference for AI runtime usage, the rise of large reasoning models like OpenAI's ChatGPT-03 and Gemini's DeepResearch, and the rise of AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based AI-based

alternatives to the transformer model for training LLMs, which was introduced originally in 2017. And number four, changes in model scaling laws that focus on inference time versus training time scaling. Sounds like reinventing the past, all four of those. My commentary, we should be careful about using the adjective bold to describe 2025 AI predictions.

As we discussed frequently here, hype about the potential for AI and AI agents will hopefully crest soon. Everything bold and provocative from the death of SAS to the rise of immortality has already been predicted.

2025 should be a year when we digest what we've been fed and shift the focus to implementing AI responsibly. Of course, we'll link to that full article in today's show notes. Now shifting to what is consistently one of the best episodes we do all year. You know today's guest as a venture capitalist, serial tech CEO and CMO, vocal advocate for social change, lover of great music, and

and Sassmetrics OG. His talks are consistently among the top few highest rated at Jason Lemkin Sassster.

I know him as a friend and the first four-time repeat guest on the podcast. Today marks the third annual Dave Kellogg Predicts the Future of Everything podcast. The original was not a predictions podcast. One thing to love about Dave is he is not shy about making bold predictions and equally not shy about harshly grading how he did the previous year.

For those who tragically don't know Dave, he's an entrepreneur in residence at Balderton Capital, whose illustrious career has spanned stints at companies like Host Analytics, Salesforce, MarkLogic, and BusinessObjects before it was acquired by SAP. And gosh, without further ado, Dave, my friend, pleasure to welcome you back to AI and the Future of Work. Catch us up. What have you been up to the last year? Oh.

Well, thanks for having me. Episode 320-something. Wow, that's amazing. Amazing success on the podcast. Last year was a busy year. I always say that demand for gray hair goes up in a downturn, right? Like when we're in a Zerp environment, you want people who don't know what they can't do, right? Because they're doing amazing things very quickly. And then when the winds change, people are like, hey, wait a minute. How do we run this stuff efficiently? How do we scale it? So I'd say for me, if I had to take one word for 2024, it was busy.

This is unscientific, but I think you did better than ever in your hit rate on 2024 predictions. Does that sound fair?

I did. It was a good year. I mean, in many ways, you could extrapolate. So, and I like extrapolation because it's reliable. But yeah, you could see, like, for example, a year of efficient growth. You can kind of see that one coming, right? It wasn't a big surprise. You knew that's what people are going to focus on. But yeah, I think I hit like eight or nine out of 10. I try to score them, you know, somewhat objectively. You know what? I bet on RAG. I love RAG. I think it's super practical. You know, I think RAG is a hit.

Some of these things, once in a while, I'm guilty of manifesting. Like I talked about SDRs, like the world finally figuring out how to use SDRs. Maybe that's a little bit of manifesting, but I do think the AI SDR has helped us realize what part of the job could be AI eyes and what part couldn't. The miss, of course, well, two things. One, I missed on option repricing.

Because it looks like that peaked in 2023. So companies were a little bit ahead of the curve there. But frankly, I wasn't sure it was even in favor again. But usually the downturn, you see repricing, as we know. But the one I think I missed on was peak podcasting. I think podcasts still have huge momentum. Here we are in a podcast behind them and are still growing. So I think I had a miss there.

I mentioned in the intro, you're a harsh grader. You do grade your own homework, but you are a harsh grader. Now, a couple of these, though, I felt like you kind of tossed yourself a couple of layups or alley-oops last year. One of them was AI climbs the hype cycle. I mean, come on. That was a gimme, right?

Well, it sounds like a gimme, but remember the hype cycle. I could have easily said AI hits the peak of inflated expectations. I didn't. I said AI was still on the way up. So it wasn't just saying that AI is popular and people are still talking about AI. It was kind of a specific reference to the hype cycle model. And are we there yet? Are we at peak? And I didn't think we would be, and I don't think we were. Maybe 2025 we will, but the prediction was for 2024, and we certainly didn't hit peak AI hype.

And the term you used was climbs. Climbs, yeah. Yeah, that was what I thought was pretty much inevitable. Climbs is doing a lot of the work there, yeah. You talked a little bit about RAG and the nature of search changing. One of the predictions from last year was beyond search. Maybe unpack that a little bit and how you've seen that evolve.

When I ran MarkLogic, people probably don't know the company, but there were two descriptions. It was kind of half database, half search engine, or what you get if you lock a search engine PhD in a database PhD in a garage for two years. Those are the two descriptions of what it was. So I got backed into knowing a lot about the search business in general. First and foremost, enterprise search, but also internet search. And

And that prediction comes from the name of an ancient search blog called Beyond Search. I can't remember the guy's name who predicted it. Oh, come on. It'll come back to me in a minute. But that guy's been writing that blog for decades. And the holy grail in search was always answers, not links.

Always. Like I could have taken you to a conference in 2004 and people were saying, we need to get to answers.links. And Stephen Arnold is his name and knew it would go back to me. And, you know, this is the world of MLSs, not MBAs, right? When you go to information retrieval and search, those are the people, some computer scientists, but you were in our customer base. We had as many MLSs as we did masters in engineering, right? People who specialize in information, organizing it, taxonomy, search, you

So to me, I felt like that prediction was, I think Stephen's finally right. Right? That we're going to move beyond search as in type in a box, get a list of links, and we're going to move more towards answers, not links, which has been the holy grail for decades. And the reason I wanted to highlight that is, unless you hung out with that particular group of people, I don't think you could have known all that.

There was only one. Remember ask.com? I think they tried to give answers instead of links, but it was kind of lame. And obviously, Google's AI summaries are answers, not links. And people have tried, but I think we're really going to get there. By the way, I think there's massive implications for marketers because now winner take all. You're either in the results or not. There's no consolation prize below the fold. Page two doesn't exist anymore.

Fast forward to one of your 2025 predictions. We can jump around a little bit. SEO is replaced by PR. What do you mean by that? So that was a crazy one, right? Because I think SEO, two things are going to happen there. So the prediction, just to be verbatim, is PR is the new SEO. I do think people should be looking at LLMO. I never want to call it chat GPTO, LLMO, whatever you want to call it. But there's a new field of optimization there.

that says, hey, instead of optimizing for being in the first set of links on a search page or for being in the featured snippet or for being in the summarize answers AI, but you used to optimize for featured snippet. Well, first link order and then featured snippets if you were good, right? And now there's a whole new game in town, which is how do you optimize for either ChatGPT or in Google just the generated AI result? So I think that's a very important part of the trend that's not in the headline.

But the funny thing is, I read... So Rand Fishkin, the founder, I think, of Moz originally, wrote a great book. I'm not going to remember the title. Oh, come on. It was like Confessions of a Founder. It was kind of a tell-all, a very kind of personal, emotional tell-all on his very mixed experience with Moz. It is a really... Lost and Founder. I knew he would get it. It's the name of his book.

A must read if you want to see beyond the kind of Tinseltown aspect of Silicon Valley, where you and I have both spent some time happily both in Tinseltown, but beyond it. Rand covers that in the book. And Rand has this explainer video that basically says, if you want to show up well in LLM results, have your brand mentioned alongside the right words and phrases in authoritative media.

keyword, authoritative media. Well, what's that? Big name publications. How do you get in those? Answer, PR. So I joked in the blog that, hey, if we live long enough, everything comes back around, right? PR was big when I started marketing in the 80s and 90s. And now here we are again after PR was basically, in my opinion, written off for dead for maybe a decade, maybe two, depends on the count. There are many approaches to LLMAO. I put some links in the predictions post.

But one of the big ones, and I found this Reddit thread as well, the Reddit thread is titled, it seems like PR is a great way to show up in AI, independent work from RANS, but they're all saying the same message. So that's where I was coming from with that. What it basically means for marketers is you better dust off your PR. If you fired your agency, maybe go get one because...

Personally, I'm very worried about LMO. And by the way, we all saw the, who was it, HubSpot, the HubSpot traffic scare. I don't know if you follow marketing Twitter that much, but that was a big thing. And I was like, oh my God, we're freaking out. The traffic's dying. Why? And is it LLMs? Yada, yada. So I think traditional search will die. Like Jason Lemkin texted me the other day basically saying, hey, look, I'm still getting plenty of inbound from Twitter. And so do I. So nothing is overnight in Silicon Valley, but I think this one's going to happen pretty quick.

Right before that one in the predictions post, number eight is unlikely revival of branding. Yeah. Unpack that one. I've got a couple specific questions based on your commentary, but why'd that make the top 10? So I'm so glad you picked that one because it's probably my favorite between us. I'll tell you where it came from. As I said in the post,

I increasingly heard marketers basically say, there's a new marketing buzzword called sea of sameness. I hadn't heard it before. And I started hearing it in 2024. And CEOs and CMOs would talk about, well, we're lost in the sea of sameness. Everything's the same. Customers can't tell the difference. And look, at some level, we've always been in the sea of sameness. When I was CMO of business subjects in the early 2000s and late 90s,

A lot of customers said, you and Cognos look kind of the same to me. And they kind of weren't wrong, right? They were kind of the same. They were both BI tools. There were certainly differences. But first, the C of sameness thing isn't a new notion. Like, are we actually differentiated and to what degree? And do you have to be a connoisseur to tell the difference, right? Like, I can tell the difference between a Pinot Noir and a Cabernet Sauvignon. Not a connoisseur, but like to tell, I don't know, to tell a 1990...

Lafitte Rothschild from 1995. Lafitte Rothschild, like 0% odds, right? None, right? So there's always differences, but can people see them? And you have to be such a connoisseur that they're lost, right? And that's this notion of the sea of sameness. And because people feel lost in the sea of sameness, it's not kind of a fashionable thing to say. People are following this logic path that goes branding is the last bastion of differentiation, right?

Like, we can't differentiate on product anymore because they're all the same. SaaS made all the products the same. Therefore, they're undifferentiated. So the last thing, the last chance for differentiation is branding. Therefore, we need to do branding. And I've heard a lot of people say that. I think it's categorically wrong, but we could talk about it more. Okay, so I'm going to call this the post-truth era. You have to question everything you see.

And in that era, first off, I also believe there's kind of a homogenization of all content. You know, everybody, all mediocre creators have access to the same tools that, you know, just reverse everything to the mean in terms of content and everything starts to sound the same. Is it harder or easier to build a brand in this era?

So it's probably harder. Ironically, you know what? I never thought of this angle, Dan. This is a little bit frying pan to fire. If you think it's hard to differentiate your product, go find out how hard it is to differentiate your company. I hadn't taken that angle on the post.

But it's a really good point, right? Anyone who knows anything about branding would quickly admit it's a multi-year exercise. It takes enormous discipline and focus and consistency of message, right? Like branding is not a little thing. It's not a one-quarter project, right? It's kind of capturing your essence, distilling it in a clever way, and then communicating it until you're bored, and then communicating it 100 times more until you're beyond bored, right? That's how you build brand.

It's very hard. So I always get nervous, to be honest. If there's a short-term problem, like we're missing numbers, we don't have enough pipeline, if branding is the answer, I am officially terrified, right? Because that's just not how branding works. It's like saying we're low on power. Let's go build a nuclear plant. Well, wait a minute. That's going to take 20 years and permits. It's just that the timeframes are not aligned. So I haven't thought of that, Dan. But yeah, I think it's this kind of post-truth environment, as you say, right?

I think branding is probably harder too. And I never really thought about it because I refuse to accept branding as the answer for a pipeline problem, I guess. To me, I put this in the post, but my daughter, I was the plugger. She works in CPG. She differentiates yogurt, right? So she gets fermented milk and she can differentiate her fermented milk from the other guys. And we can't differentiate our SaaS financial planning app. Come on, right? Like we've got it easy compared to the CPG guy. Yeah, well said.

Now, a cynical commenter on the Predictions blog post mentioned that we're the database predictions in the annual Predictions post from Kellblog. And the reason I just teased that one out is because most of the time your predictions include things that are quite technical, quite nuanced, oftentimes related to database architectures. If I were to say there's a theme in this year's predictions,

There's more social commentary. We talk about, you know, we've already mentioned, you know, PR, SEO, branding. Is it decidedly, if you just look at the arc of your predictions, is it decidedly different tone? Is that intentional or why did you choose not to focus on like the deep tech, the infrastructure, like some of the things I mentioned in the fun fact, you know, the venture beat predictions were all, you know, about the future.

the foundation models, the transformer architecture, things like that. Was that something you were intentional about? So the easy answer is not intentional. The trickier answer is why. And I think there's three reasons why, some of which might be interesting. So let's just talk about them real quickly. I am a lover of databases. There's plenty of database books in the cabinet. I still read database books.

I'm probably going to say something stupid that I'm going to miss here, but I've gotten a little jaded about special purpose databases because the general purpose DBMS guys are so good at making them features. They are so, so good at just absorbing anything in as a feature, whether it be text search, whether it be column-oriented and OLAP.

That used to be a special class of database, Redbrick or S-Base. MicroStrategy originally was a relational OLAP database. Absorbed. Column orientation, Vertigo, specialized database. Absorbed. Vector databases, the most recent innovation I can think of, I think, being absorbed. So I feel like special purpose databases end up pretty niche and complex.

They basically have one of two outcomes. They either end up niche because they're not worth absorbing and they're very good at doing something that not many people do, or they get absorbed. And I've yet to see, you know, the closest thing I can think of to a database disruption. In my opinion, NoSQL kind of failed. Many of the architectural patterns survived, like massive parallelism, handling unstructured data, schema independence, or at least schema liberty, right? Loose schemas. That all survived. But

I don't know. You know, the biggest thing I think that has changed databases is Spark, but Spark's not a database, right? But it's certainly changed analytical processing. If I had to make three predictions on the fly, I'd just tell you two of them right now. One would be, I think ETL gets disrupted by kind of co-pilots and AI agents.

Because ETL is a space where there's lots of relatively low-skilled programmers writing lots of relatively basic scripts to move stuff around, concatenate these two columns, map this to that. It's not glory work at all. So I think AI, either through agents or through co-pilots, a co-pilot can just write the code for you, or maybe an AI agent can just do the whole task for you. So I think there's disruption coming to ETL land. So that's in database land.

Data catalogs, I'm not sure about. I've done a lot of work with data catalogs, and I'm a little bit confused about what happens to them. One company I'm working with is doing something really interesting, I think, which is kind of plug-compatible Spark. So Spark has been enormously successful, but it has performance issues. And these people have basically rewritten Spark, and they have a higher-performing, faster-performing

plug-compatible version. So we'll see. It's highly disruptive, but that to me is probably the most interesting thing I've seen is, wait a minute, somebody's going after Spark and they're not doing it but trying to make a better Spark. They're trying to make Spark better. And that's a twist. The company's called Lakesail. Very, very early, but I found what they were doing super interesting.

He went PhD level on us there because I thought Spark was a database. I don't think it is one. These are semantics. Yeah, at least the data store. And I certainly, I would consider NoSQL to be a big success. Look at Mongo and many others. And I thought that the vector database is really the first kind of niche database that's a standalone database.

piece of technology. But maybe I'm wrong on all three. I would say, look, Mongo's been very successful. I was an advisor to Mongo. I love the company. I love the product. In many ways, it was kind of an open source and better version of MarkLogic. So I have zero problem with Mongo. Maybe I'm just disappointed because everybody always promises, oh, we're going to have a database that's got relaxed schema. You can put in unstrutured data. Things don't have to be in rows and columns. And it's going to replace everything.

And it doesn't replace everything. It replaces the stuff that shouldn't have been mapped to rows and columns. Right? That's all. So part of me is burned out on the hype. I think Mongo has two big advantages. One, developers like it. So I always say, if developers got to pick what databases we use, we all would have been using object databases. Right? Back in the 90s, C++ was blooming. Object databases are perfect interfaces to C++, to languages, object-oriented languages. But they didn't meet the database requirements of the DBAs.

I feel like Mongo took advantage of that. It was kind of revenge of the object database. Hey, developers like it. And the other thing was that like MarkLogic, you did not have to have a schema, right? In most databases, step one to find a schema in relation to the database is a table and then stamp rows in, right? That,

back to the schema. And that's not true in documentary to databases. So that's what I was trying to say. I forgot about Mongo when I was talking. A lot of the concepts like relaxed schemas definitely succeeded. But, you know, column orientation succeeded. I don't think Vertica is a particularly big business right now. I don't know for sure. By the way, my favorite tagline ever, remember Vertica's tagline, trivia, database trivia time.

Best tagline ever. The tables have turned. What a fantastic tagline. Wow. That's as nerdy as it gets. Brilliant. Now I have to dust that one off.

So when I mentioned that topic about less kind of infrastructure, deep tech, database-related predictions, I was reading between the lines of the predictions, and I thought what you might say was, unless these were weekly predictions predicting anything related, certainly related to AI data production,

New software patterns, architectures, it's just not something you can ever do again on the order of years. But maybe I read into your post too much. Yeah, unfortunately, I think...

I think it was more that just happened to be working with fewer database companies, frankly. Some of it's my client mix who I'm hanging out with. And I got really excited about vector databases. So you did hear a little bit of jadedness there. Because when I first heard of them, I'm like, ooh, new database type. I love new databases. Hey, I'm doing some work with Influx time series data.

Great company. Love what they're doing. Really good at time series, right? And you could extrapolate. It's good at sensor data, right? It's going to be good at real-time data, right? Anything where a data point isn't a row but almost a vector, right? Like, oh, we don't have one data point for the temperature today. We have 1,000 samples per second of the temperature on the inside of a volcano, right? And in a relational database, in a regular schema, every...

you'd be adding 1,000 rows a second for that one sensor, right? It just doesn't work, right? So time series databases work for real-time apps. Vector databases work for basically where you're storing vectors, as you often do in AI applications and clustering. But I just didn't find any general-purpose database that got me razzed up. Of course, I'm going to forget one. I'm going to get a hate mail after this comes out. But...

But yeah, I just haven't bumped into anybody. The coolest thing I saw was these Lakesail guys. I'm working with another interesting company called Prophecy. And Prophecy is trying to disrupt ETL. They have a co-pilot now. They're doing more work beyond that. And I like that space. There's a space desperately in need of disruption. Informatica still rules. Somebody needs to come along and disrupt them.

Yeah, well said. So you mentioned in the opener, you're now an investor at Balder Capital, VC firm based in Europe. You're a Silicon Valley guy. And...

Puts you in the catbird seat to answer this one. So last year, 2024, for the full year, somewhere around 6x the investment in AI-first companies went to venture-backed companies in the U.S. versus Europe. Somewhere around 100 billion versus around 16, 17 billion in Europe. Any thoughts on, will that trend change in 2025?

So a couple of things, Dan. One, just to slightly correct the record, I do work at Balderton Capital. I'm an EIR there. And my job is actually helping portfolio companies doing all the stuff you'd expect me to do. Go-to-market, strategy, positioning, marketing, Salesforce, how to build a Salesforce, metrics, all that stuff. So I don't really work on the investment side at all. In fact, Balderton companies are often shocked by how ignorant I am because it's like, hey, I wasn't on the investment community. I didn't read the memo. I'm just here to help you build your go-to-market. Let's talk about your sales VP.

That said, I do spend a lot of time in Europe, as you noted. Personally, look, I think Europe has an enormous amount of tech talent. Part one, the educational system is excellent. They produce lots of great engineers. And for whatever it's worth, the French were ahead of us by decades in AI. I don't know if you remember back in the 90s, like every French engineer's degree was in AI. And there wasn't enough data to analyze, and the models weren't good enough, and you didn't have to compute. But...

They have great technologists. I'm on the board of a company called TechWolf, three guys who met getting their master's degree at a university in Belgium. You know TechWolf. Those guys are serious technologists who kind of happened to back into an HR use case. So the number one thing, I'm going to be very American. I worry about regulation. And I'm generally not a...

I don't know if you saw the thing over the weekend, but there's some cooler that cuts people's fingertips off when they slam the lid. And it's like, we really should have consumer protection because I'd still like to be able to type on Monday after I go to the lake. So I'm okay with that. But I do worry that you got to be very careful with AI regulation. And I think we get a bit of a... Look, here's the interesting thing. Do we get a competitive advantage because we're unregulated relatively? Or do they get a...

We can argue in data privacy, they're ahead of us because they had to do GDPR and they had to do it early. Calibra exists. A multi-billion dollar market cap company, their initial use case was all GDPR. And very few American companies actually really understood it. So if you had to comply, and it's not just for European companies, it's U.S. companies working in Europe, that Calibra was a good example of what to do.

So you can look at both sides of this. When you lay the regulations down as constraints, like they've done with privacy in Europe, where maybe they were actually ahead of us. Like, I'm not feeling great about data privacy right now in the US. How many how many mails a week do you get about data breaches? Right. Literally, it's mails per week at this point for me, let alone Musk and Co. We won't go there. The biggest data breach of them all.

Right. I'm not feeling great about data privacy here. Maybe they were right on that one. There wasn't as much screaming, to be honest. Now there's much more screaming from Silicon Valley about how regulations are going to cripple AI. But obviously, you've got Anthropic. You've got some big leaders. Balderton is a super interesting company called Rider, which we'll hopefully talk about later. When you talk about non-obvious things in AI, I think what they're doing is non-obvious. So in terms of the investment, I don't know. I don't have the numbers tip of my head. I mean, U.S. investment leads European by a ton anyway. And so not shocking, I guess, is the short answer. Yeah.

Does the European tech community look at the U.S. right now? Take this in whatever direction you want, but is the U.S. the laughingstock of the world or are we the mouthpiece for the world?

I don't know. I'll just do anecdotally. I do get a lot of, are you okay? If I had to summarize it all, I would say it's a little bit of concern. Like what the hell is going on there? Do you ever answer that truthfully? We'll come back to the answer in a second. But the other, I just have to share this joke because it's too funny. You know what Robin Williams always said about Canada? Do you know this one?

No, I don't believe I do. It's like living in an apartment above a meth lab. I think I get a lot of that vibe from people. I think they're concerned. I think they're worried about what's happening here. I think the VCs here have done something unprecedented, which is start talking about politics and very pro-American. I don't know how you raise money in Europe.

Sorry, not race, but how you invest money in Europe. Like what European founder wants you if you're pro-American, America first VC. It's like, wait a minute, I'm a European and I'm trying to expand into the US. So I do think there'll be long-term consequences for these things, but people don't talk about them that much. As you know, Dan, the world we grew up in, you didn't talk about this stuff.

At all. And now you've got Andreessen and you've got Sean McGuire, Sequoia, and you have the All In brothers. You have a lot of people very vocal about politics and public policy. And I think it's going to be a couple of years before we see the fallout from that. I will tell you a great soundbite I heard this morning talking to a European who basically said, wow,

The unacceptable has become fashionable, which I thought was the greatest summary. And this is a person who's been around as long as we have. And so they know the history. And like talking about this stuff was absolutely unacceptable in the 90s. And now it's like super fashionable. Oh, Marc Andreessen is talking about how to run a country and, you know, et cetera. So but yeah, I think the short summary is they're a little concerned about us and perhaps rightfully so, you know.

The script totally flipped. It used to be not that we were being muzzled, we just weren't asked and we didn't choose to share those perspectives. And now the opposite is normalized. The act of not sharing your views is perceived as, you know, everyone wants to read into what you think and what your opinions are. Whether you vocalize them or not, you're making a statement.

So in my opinion, so true. And I'll just say, like, Mark Lodrick was backed by Sequoia, and that was in the early 2000s. I was CEO there from 2004 to 2010. So I got to know them at the time. And I would just say that they tried to stay out of the press. Like, just a classic old-school VC. Like, during that era, they weren't trying to get headlines. They would talk about their big successes, obviously. They'd do speeches once in a while. But...

They weren't trying to be in the press. They weren't trying to, it was more like quiet competence. Back in the old days,

They used to not mark up their investments in the old days. They would say there's no reason marking things to market because it'll just be misleading. You invested a million dollars. We'll tell you when we sell the company what it's worth, and we'll tell you about historical IRR, but we're not going to play the market. I think they're forced by their auditors to do mark to market, right? So we're talking about a very conservative philosophy. I may have screwed up the details, but the philosophy was, let's just go be super competent, find the best investors, not to talk too much about ourselves,

right and now to this is another favorite quote but you know how people call the harvard a uh a hedge fund with a university attached right like like andreessen horowitz is a publishing company with a venture capital firm attached right this is what it feels like to me like the complete inversion um so some people are dying to be in the headlines all the time you had some choice words about the world's favorite business water cooler called linkedin

How does that weave into... I mean, I thought it was a broader theme. You had a very clear perspective on that. Not just the prediction, but...

What's the genesis of that? Well, first off, summarize your thoughts. And then what was the genesis of that? Yeah, so the thoughts, this would be another example of manifestation. I'm actually not, I'm trying to manifest the opposite of my prediction in this case. But the prediction is LinkedIn enters the social media death cycle. For various reasons that you can guess, I have decided, it took me a couple of iterations to get off Twitter because I just, A, I couldn't support what was happening there. And B, just my experience was more and more garbage.

And again, I had a debate with Jason Lemkin about this just the other day on Twitter, ironically, as I was leaving. But he's like, if you have all the right mute words and you don't engage, you can still have a really good experience. Roughly what he said, roughly. And I'm like, you got more discipline than me because when I'm on the highway, if there's a car wreck, I slow down and look. So I take the bait. So I can't go there anymore because I have lots and lots of mute words.

But in the trending topics, the right list, they get me. And you can say it's a Dave problem, not a Twitter problem. But the net result is I literally can't go there. I'd be like putting an alcoholic in a bar or something. It's like, don't put me in Twitter because I am going to go look at all the stuff that drives me crazy. So as a result of that, I moved to LinkedIn. And I said, ah, LinkedIn can be my new home. And I immediately found

a lot of problems with LinkedIn. But of late, the thing that was driving me crazy, I actually put a specific post up there where Andy Price, the kind of famous recruiter, right, from Schreikler Price, yada, yada, now his own firm, Artisanal Ventures and Artisanal Search,

The thing I noticed in the post was he put a comment up and the notification I got was Andy Price and others have added 34,000 comments to a post. I'm like, holy cow, I've never seen that on LinkedIn ever. Have you ever seen a post that had 34,000 comments? And I look and it's a post about Israel and Middle East politics. And I'm like, oh, gosh, wow.

I'm trying to get away from this stuff. And for me, people can say I let politics bleed into my feed. Like I certainly do on Blue Sky because I have views and I put in my fact what I talk about where. And if you want to know my views on politics, go to Blue Sky. I do not on LinkedIn. I'll do an analysis of messaging, which I did do. But to me, that's a marketing course using politics as an example. I don't care who you vote for. Let's look at what the Republicans did. Let's look at what the Democrats did. Here's what they said. Whose message is better? I don't care about the content.

I care about the marketing content. I don't care about the political content. So what I noticed on LinkedIn was just more and more politics coasts breaking through and getting extremely high engagement. And because you also know I'm a metrics guy,

I'm like, you measure what you manage, yada, yada. It's like, okay, these guys view success as engagement, and that's the path to hell. Because these things get engagement. They just take you up to the point where everyone leaves your site because they can't stand it anymore. I think this process... I can't remember who coined the phrase. It's somewhere on the blog. In shitification. And I think it's the perfect phrase to describe what happens and how...

I think LinkedIn, unfortunately, is going to, Cory Doctorow said it, is going to get inshitified. And I think, frankly, they're taking half measures against it. The next part of the story, sorry to keep talking so much, was I then found a feed preference on LinkedIn that said show political content and feed on off. I turned it off.

I said, I don't want to see this stuff. And it didn't work. I kept getting it. And I'm like, okay, these kind of half-hearted measures, you might be able to control this with a feed setting, but it has to work. LinkedIn has no easy way to block an author, by the way, to mute. You can only block. But you can't mute somebody. I've never found a way to do mute words. You can see not interested.

But on a repost, I'm never sure if I'm saying I'm not interested in the reposter or the original poster. So the semantics of the operation are ill-defined. I do it anyway. But I'm like, I wonder if I'm never going to hear from Fred again because I'm saying not interested. So they have what I would consider to be half-hearted measures. And just given the polarization and the engagement,

I mean, if they let engagement drive the bus, we know where the bus is going, right? Off the cliff. I don't use Facebook anymore. I don't use Instagram. I use Instagram now, but just to watch fly fishing videos, like I never see my friends on it anymore. Right. So we know the path. You get lots and lots of users, but...

But anyway, yeah, that's what drove it. As you tell them, passionate about it because I'm trying to, I like social networking. I like talking to people about SDRs and, you know, metrics. And I used to do that in Twitter. Can't do it. Facebook never did it. Instagram never did it. Blue Sky, I'm trying. LinkedIn, it's hanging on, but I'm worried about it. I was shocked in the 24 hours after the deep fake news, deep seek, that

no pun intended, after the deep seek news dropped at how many experts in my network there are on geopolitics. Yeah. Oh yeah. Yeah. Yeah. Or the, the Musk salute was another big one. Right. And I just don't want to see that on Twitter. That's not why I'm here, but,

My joke is, you know, like there used to be a dress code at the red carpet club back in the day in United. They're like, I'm sorry, sir, this is the business lounge, right? That's what I feel like saying to people when you're on Twitter. No, this is the business lounge. Get your kid out of here who's screaming and get your politics out of here. Like we're here to talk about business. And right now they're not. It's bleeding in and they're not stopping it. This one's adjacent to your predictions, but I always got to get in at least one metrics question while you're here. Selfishly, have we seen the end of the rule of 40?

Is that something that we're going to talk about in 2026? Is that we're going to bemoan the death of any company that was efficient enough to achieve that? And maybe just for those not familiar with it, maybe just... So the Rule 40 is a kind of SaaS metric that basically says, if you add your revenue growth to your profit...

it should be around 40. And you can define revenue as either ARR growth or subscription revenue growth, or there's a number of ways to define revenue, usually ARR revenue. And you can define profit as EBITDA, adjusted EBITDA, free cash flow. I'd say the standard is ARR and free cash flow for what you measure, especially for private companies. I think the rule of 40 is here to stay, at least for a while, is the short answer. Because I like its intent. I like its simplicity.

You do regressions in companies that are compliant or worth more than the ones that aren't. And it's just a very elegant way of saying, I need you to balance these two things. So, you know, if great, you want to grow 100%, then you can lose no more than 60. It's a highly scalable rule, right? You want to grow 200%, rock and roll, go at it, right? You want to grow 10%, then I better see 30% of the margins.

Because we're certainly not going to get a revenue-based valuation on this company. And if we're going to sell this company for 15 times EBITDA, right? The companies that make me sad are the ones growing 10% at breakeven. Because it's like, you're not going to get a revenue multiple off 10% growth. And 15 times zero is zero. So you have a business that's literally worth nothing. So you have to do something. So I like it.

I would say there was a new rule proposed last year by Bessemer called Rule of X, which is kind of a growth-weighted Rule of 40. Because I think the one Achilles heel of the Rule of 40 is a point...

of growth is not really the same as a point of profit. And if you do all the regressions, it turns out it's worth, in today's environment, about 2.3x. So it's kind of a 2.3x growth-weighted rule of 40. And I think that is the one improvement you can make because growth compounds, profit doesn't. So

I like the rule of X. I think it's a terrible name. I think as defined, X gets redefined. There's certain flaws that are more technical in how it's defined, like how often do you reset that weighting. But in any case, I like it. Basically, I would say a 2X growth-weighted rule of 40 is a fantastic metric, and I think it's going to be here for a while because I think we're a long time, Dan, before we're back to Zerp and free money. Now, there are always exceptions. In AI, it's Zerpville, right? There is free money.

By the way, I just watched a company raise money at 40x revenue. So the insanity is not completely over. I would say it's just highly concentrated on companies that either have exceptional numbers or companies in AI and then take the rule of 40 and throw it out the window.

capital still it's still zerp it's still 1999 capital's free grow as fast as you want but uh i do feel like if you see it's coming more and have a have and have not thing i don't know if you've seen that but the haves can raise money at insane valuations they have not literally almost can't raise it at all um and that is a trend i don't particularly like i hope you're right that the rule hasn't gone away i found it uh i hope the zerp era punch drunk mentality doesn't

doesn't come back but as the ceo of a fast-growing software company in ai i i'm glad that you think that it can still be achieved it is a hard metric but uh you know i'd like to think there's still value and discipline absolutely and for what it's worth i do think ai driven go-to-market tools are going to make look look ai sales and marketing costs about 2x r&d in most sas companies even a kind of semi-mature sas company kind of 20 million and up

And I do think AI can help reduce the sales and marketing percent. So I'm optimistic on the profit side of the rule of 40 as well as the growth. Dave, sadly, this happens every time we chat. We're about out of time. In fact, we're way over time as always. But it's always worth it. But I'm also contractually obligated to not let you off the hot seat without asking you a question about music. Uh-oh. You know it was coming, but...

Turns out at the 2024 Den Company residency at the Sphere, the single longest song was a 14 minute and 38 second St. Stephen. I don't believe it was the, I was at one of the shows last year. Are you taking the over or the under on that one? Is the longest song that they play at the 2025 residency going to be more or less than 14 minutes and 38 seconds?

Well, I mean, speaking on behalf of fellow Grateful Dead fans, I certainly hope it's longer than 14 minutes. That used to be a short song, right? Back in the day. Not quite a first set song, but pretty short. I'm going to hit you with one piece of Dead trivia since we're doing this. This is kind of fun for music people in general. Do you know why the song called The Eleven is called The Eleven?

I don't, I should, but I don't 11 beats per measure. It's actually the time signature. So if you listen to the 11 and it goes 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 2, 3. So it's actually the time. So my brother actually figured it out by listening. He's a magician and he's not a deadhead. I took him to a show and he's like, Hey, what was that song? That was an 11, four time signature. I can't, maybe it's 11, eight, but what was the head of 11 beats per measure? And I was like, Jesus, Fred, how did you notice?

Well, if you and I don't get a chance to hang out in Silicon Valley, we'll definitely be seeing each other at the Sphere sometime soon. I'll be there for five shows this spring, so hopefully I'll bump into you. All right, we'll make it happen. Hey, Dave, where can the listeners learn more about the good work that you do? And of course, subscribe to Kellblog if they're not already.

Don't miss our podcast. It's a little bit geekier, but SaaS Talk with the Metrics Brothers. We'll literally talk for half an hour about a single SaaS metric, like implied ARR or CPP, payback period. So you have to be a true aficionado, but check it out as well. Where is that at? It's called? Apple Podcasts called SaaS Talk. It's like Car Talk was the inspiration. Okay.

I hope you have the Tappet brothers on there occasionally. We do. We're clicking clack. Sorry, they're clicking clack. We're growth and cack. Kind of corny. Okay. Kelblog is at kelblog.com, right? Correct, yeah. K-E-L-L, like Dave's name, blog. Dave, always a pleasure. Looking forward to this already. I got the date circled for you next year, right? Come back often. Fantastic. Thanks for having me. Always a pleasure. Can't believe we ate all the time already, so it's great talking to you.

Great work as always. Gosh, that's all the time we have for this week on AI and the Future of Work. As always, I'm your host, Dan Turchin from PeopleRain. And of course, we are back next week with another fascinating guest.