Navigating Growth in a Changing Environment | Paul Lippe on BeyondSaaS Ep 006

Well, I've been around this topic for quite a long time. I think it's probably the case that for most companies, if they're selling to large enterprises with IT, almost everything that's become the norm around growing SaaS companies is going to change in the next 24 months. So if you're running the 2014 playbook, you might want to substantially revisit what you think is the norm.

My golden nugget is somewhat random comparatively speaking. I've spent the last two weeks tracking my time in a spreadsheet every moment just to see how I could be more effective at doing things. And I kept a column on the right of what I can get rid of. And I found probably at least hours a week of things that I can hand off, which is pretty substantial. So it wasn't clear before, it's clear now, and that's just kind of a fun exercise.

I probably had the same idea for at least 30 years, which is most business documents, most business language is kind of paint by number. I'm a lawyer, but I was the head of business development at Synopsys, which is probably the world's biggest actual AI company and the world's biggest company that's built around a structured language for chip design called VHDL or Verilog, and then also now it's probably the fourth or fifth biggest software tools company as well. Whether you're talking about a contract or a policy or an RFP response or your earnings release FAQs, most of what you say is what you said before, tweaked and configured for a slightly different audience, slightly different customer. And the challenge is largely, who has to sign off on it? And so by and large, the problem is latency. It just takes a long time to get things out the door and that slows down cashflow, makes things more complicated and actually distracts or detracts from people really optimizing what they're doing. So obviously there's a lot of changes in the world in the last 25 years. Gen AI, you know, it's a bit like 1781 somebody invents a steam engine to take water out of coal mines. Okay, what are we doing with that thing? You know, so right now there's a lot of thrashing. You know, we think our thrashing is the correct one. I guess we'll find out.

You know, like anybody, I'm trying to scratch my own itch, right? So I've been around lawyers a long time, kind of goofy, make things more complicated than they need to, often don't focus on what's important or most important. Other knowledge workers in the enterprise are again, you know, more like lawyers than you think because most of the knowledge worker functions are still rooted in the professional culture of their field, whether it's marketing or HR or procurement. So they haven't really adapted to a scale, a systemic kind of orientation. If you think about most things that are documents, if you're designing a chip, you might have 500 people collaborate on the same chip, right? It has to be portable, has to be testable, all this stuff. Most documents probably have three or four people that collaborate on them. Yet the document itself becomes an essence code or a set of protocols because it defines things that the company is going to. I don't want to see Gen AI eliminate knowledge worker jobs. I think if anything, the net effect may be we re-shore a lot of jobs that got off-shored. And so U.S. employment hopefully will be up. To that point, I saw a really interesting comparison that when photography came along, all the artists were mad that the photography was going to take over everything they did, and now painters are more valuable because it's human effort, human work and their artistic translation, and we have both, right? So that's kind of how I see it. I don't know how these things will play out. If they'd called it Gen Autocomplete or Gen Spellcheck, everybody would be a lot calmer. Because it was called Gen AI, people have sort of imputed 40 years of wish fulfillment about what AI may or may not be. What it is specifically is a way to understand language and generate language faster. A lot of the time knowledge workers spend is just organizing stuff and checking stuff and whatever. My task of driving from here to Los Angeles is easier if I use Google Maps than if I use Rand McNally or I use nothing. So, you know, that's our goal is just to simplify the baseline stuff so that you can focus on, you know, where the value is for the organization and the individual.

My old company was kitty corner across the street from Netscape in the 90s. And so I had a pretty front row seat in Mountain View. I lived in Los Altos. Pretty front row seat on the internet. This is way more jumbly and goofy than that was because you really didn't have the machinery of hype to the degree that we have today, thanks to the internet. You didn't have the capital availability that we have today, although at the time it seemed like massive capital. And there wasn't a tremendous amount of expectation about what this thing might be. It was viewed as the hypiest thing in the history of the universe, right? So the hype is so far ahead of the reality. And a lot of it's simply financially driven, right? So if there's capital available to build data centers, it drives demand for infrastructure. So the economics of some things are there. But right now, there are no real meaningful success cases around Gen AI other than software development and advanced help desk, which were highly automated before. So it's just so thrashy and you've got people all over the spectrum in terms of extreme skepticism to extreme big hat, no cattle, hucksterism. And so we'd like to think, do we actually know what's going on and we can deliver value? And so there's so much looking to coalesce on things that actually work. I think we just have to kind of be in the right place at the right time, deliver stuff that works, and then we'll be good. It's very different from the canonical SaaS, 3-3-2-2, inside sales, boom, boom, boom. That whole playbook, like I said, I think for most people that's dead or will be clear to be dead. Obviously, the focus on product, the focus on the customer, the focus on the user experience, those are, you know, those are timeless. But things that drove the kind of venture we got, we got inbound paying from a venture firm I knew it was going to be a waste of time. I talked to him yesterday, you know, basically recited the cookie cutter approach. I'm like, this isn't you know, this isn't gonna work anymore. Yeah, well, you know, that's how we raise money. So there's a lot of latency in these things, in the signal of what's really working, what's not. So I think this is more confused and ultimately more dynamic moment than 1997 or wherever you want to suggest was kind of the ur-year of the internet. And so you gotta pay attention and hope you get lucky with the people you're talking to.

I look at a lot of what people have sold so far. Again, differentiate between the consumer and business, right? So if I'm Google, I'm Facebook, I got Paul Lippe's eyeballs for 23 minutes and 17 seconds a day. I'm making $1.43 on serving of ads. If I can use Gen AI to get his eyeballs for 24 minutes and 41 seconds and sell a $1.72 worth of ads, that's great. I can build a big data center in Mississippi and invest in models, right? That's one business model. But if I'm a bank and I've got to actually make more money, I'm a telco, I'm an entertainment company, you know, I can't do things that are that speculative. I've got to actually have users and whatever. But what's actually happened on the business side so far is all the spending has been speculative. It's been spending on models and almost everything. If you actually asked any company, they might be spending on infrastructure. They would say everything they've done is an experiment. And again, if all these experiments were working, you'd be hearing about them. So the good news is I think our experiments, which some of them are actually not called experiments, but our initiatives, are going to work. And so, you know, can we grow 30x in a year? Theoretically no, but practically perhaps, maybe it's a hundred, maybe it's maybe six. So yeah, we're in a world where theory is inverted and so reality is up for grabs. So we'll see.

Specifically, we've always had a function which we call language engineering, which is built on what the application engineering function would be in a technical software or technical company. But I think everybody we've gotten knows is lawyers, which turn out to be great. And so they work with customers to tag data, to set up taxonomies, to set up workflows, etc. You know, the DevOps team, you could scale the business 20x with twice as many people. Sales, you know, will scale linearly. We're part of the Microsoft Azure Marketplace, so we have tremendous leverage there. And Microsoft's a fantastic partner in all this. I got off the phone with a Microsoft salesperson. So the sales side is pretty easy. And of course, everybody is connected. You know, practically speaking, if you were just doing your planning, the question is, can you hire and train the language engineering function fast enough for the customer opportunity? We've been thinking about that for a while, so the answer is we'll try. We'll see.

We're focused on say Global 2000, not China, heavily interested in entertainment, utilities, telecom, financial services. So it's easier for us to virally extend within an industry group because they tend to know each other. And then in almost all industries, companies have anointed some kind of Gen AI task force, often but not always reporting to the CIO. Those folks have kind of come full circle. So they probably thought 24 months ago this would be a little more straightforward than it's been. We probably would have thought hard for a small vendor like us to get the attention of those folks. Now we've got lots and lots of folks to talk to and lots of recognition that we've got a good solution or a good starting solution.

As they say, we don't disclose that, we're small but mighty.

There's a lot of dimensionality to how enterprises do contracts and how it's evolved and the way it is and whatever. But I think people have recognized for 15 years, maybe longer, that there's a problem there. And the posited solution was the contract lifecycle management applications, most of which are CLMs. I saw the very first CLMs at probably at Cisco 25 years ago, and maybe IBM. And for a variety of reasons, those things haven't worked. And more and more money is coming into them, AI, yada, yada. I think the whole CLM category will simply be subsumed in the broader language stack for the enterprise.

By any measure, what does it cost to do them? Are they faster? Are they better understood? Is the business happier? Is your customer counterparty happier? There's zero improvement in contracts despite the investment in CLM. And arguably, they're somewhat worse than they were. And so we're finding that that's an area that people are very interested, as a natural area to explore AI. The other thing is you don't need to scan the whole web to do basic work. Verizon speaks Verizon, Netflix speaks Netflix, so I think most companies have figured out or will figure out that the focus of their Gen AI activity should be within the firewall, within the data center. That's how they'll execute better. That's how they get competitive advantage. That's one of the reasons I think we can work well with Microsoft, that fits well with them. So far, it seems like we've sort of stumbled into the right place, the right time.

You mean on the SaaS side?

Well, I mean, if you look at the SaaS playbook, you know, 3-3-2-2 growth, sell the same thing over and over again, find your call point, got to be subscription, shouldn't be services. I think most substantial companies with IT departments will figure out that if I've got a SaaS vendor with my travel planning, I got a SaaS vendor with my contracts, I got a SaaS vendor with my IR, all the language that I use essentially reflects my business rules, my process. And so they're going to take over more and more of those rules. And so maybe I can cost reduce from 11 to four people. Then I'm hollowing myself out. So the first order, the company looks at these things and says, well, I'm worried about InfoSec. I'm worried about the cost of models. And those are legitimate, but ultimately solvable. And as you know, IT in general has not been a fan of discrete SaaS forever, right? They don't want the fragmentation, right? There's always a tension between function-specific software and broader tools, and usually the function specific wins because they can solve the problem better. But at least for the class of stuff that we're dealing with, all language is kind of created in the same way and ultimately used in similar ways. So I think companies will realize, hey, we should capture the value ourselves rather than hand it off to these 200 vendors, which will then, you know, effectively, whatever you put in, they're not going to give away your confidential information, but they will capture your know-how to some degree, and then that'll be available to your competitors, right? So there are a lot of reasons why IT and finance recognize that a fully SaaSified Gen AI, SaaS in the sense of discrete functional application, not just in the sense of hosted, that that's actually strategically weakens them.

For the vendor, the people selling software?

Satya Nadella, who's the CEO of Microsoft, obviously a smart guy. He's come out and said that the SaaS is — it's hard to find exactly what he said in a couple of different places — SaaS is dead, and these are CRUD applications. So first order, I think everybody recognized from the get-go that in a Gen AI world, you'd probably have natural language interfaces rather than structured interfaces, right? So that's a big part of the SaaS application is having a nice UX. But it's a specific UX for this specific application, right? Then the second thing is what data is there, right? Where is the data? And is the data structured, and etc. So I can barely figure out what we're doing. I can't tell anybody else what to do. I think the first order challenge is getting whatever data exists today into super well structured, what we would call knowledge sets, that therefore is going to yield good results, right? Because you can get a nice eloquent, fluent thing written on ChatGPT or Perplexity or Claude, whatever, but that's not going to be good enough for doing real business, right? So some of these applications, I guess, a lot of these rules never change and a lot of them change. So I think there's been a lot of people that have started SaaS companies without any real domain knowledge. This is called the kind of Y Combinator mindset that, you know, we'll write some code, we'll go talk to some people, we'll find their pain point, and then we'll A/B test it until we get it right, right? And that's, you know, it's super admirable from an entrepreneurial perspective. It's worked for lots of companies. I think it's going to be very hard to succeed in many, not necessarily all, of these things without real domain knowledge. Because the way people work is more nuanced and complex than you imagine if you haven't seen how they actually work.

Well, we actually got marketed to by a lead gen company. I won't say their name, although I probably should for their benefit, but I don't want them to get too busy. And they've been great. I probably get five lead gen things a day on LinkedIn. Just first order, LinkedIn has become so spammy. If you're going to contact somebody, at least make an effort to customize your message. And then when I respond, which I do sometimes out of courtesy or there's some commonality, then most people don't even follow up. And then I say, please switch it to email. And they don't switch it. So if you're going to do outbound stuff to humans, try to be a human.

That's above my pay grade. I don't know. Somebody from Microsoft said, you guys are really good at brother-in-law and roommates selling. We have a call tomorrow with a Fortune 50 company, an SVP there, and we were talking to actually a Microsoft guy. He said, man, that's great. We never can call on somebody that high. How did you get that meeting? You know, just a friend of a friend. We're old, you know, everybody we know is old. I think so much confusion and so much of these, we'll call them markets or customer sets, whatever. The fundamental issue is cultural in terms of adoption. Again, that's obvious with the lawyers, but it's true with most of other functions. There's going to be a lot of noise, and then they'll tend to be fairly winner take all, because whoever is acceptable to the culture, you know, will kind of win the whole thing. Obviously, you know, we have a candidate for who we think will be that. So I think you gotta, to the best of your ability, come in as a trusted insider and somebody who can speak the lingo and all that stuff. And I think frankly, if you look at the, you know, let's just call them the elite consulting firms, that's been their game, but because they haven't really brought anything of consequence to this Gen AI conversation, I think they're losing some credibility. I've probably heard that echoed six or eight times this week so far. So you can't just pretend you're in the club and you got to actually bring something of value. And in this super cluttered environment, I don't know if that means you just have to do one thing that's not ridiculous that's good enough, or you got to have the whole thing figured out. Who can say?

Yeah, I mean, again, if you go back to say 1997 or whatever, right? Domain knowledge that Netscape had, right? They had a build it and they will come. The browser can give people a way to navigate the web. And, you know, super successful company, but by the same token, they never figured out their business model, right? Google came along. I was actually running a company at Stanford, third of a mile from Google at the time, from the Google people. Obviously, they did a lot better than me. And they had no domain knowledge, but they had a great technology that was the right technology. And then, you know, basically they stumbled into the advertising model, right? Their first order of thought was make it as simple as possible, which was brilliant. And then they, I mean, you read this stuff, they kind of lucked into the advertising model and then everything comes from that, right? And then all the SaaS companies, some of them have been people with domain knowledge, some haven't. If it's artificial intelligence or whatever you want to call it, the operative word is intelligence, right? The operative word isn't artificial. So somehow you've got to understand the intelligence of what's going on.

Well, we were super lucky. There's probably five or six people that were already in our network or network adjacent who have proven to be precisely correct about what was going on and what wasn't, and maybe another 100 people that we know who proven to be wrong. And so we were lucky enough to be in conversation with the people who were right and lucky enough to dial up their input versus others. So I think that somehow you've got to be exposed to as many points of view, as many smart people as possible, and then do your best to filter. Of course, you know, don't have tons of confirmation bias and just look for people to reaffirm where you're already at. But, you know, do your best to filter or dial in on the people who happen to be right.

I'm on LinkedIn, but if you send me a thing on LinkedIn, please be clear what you follow up. I'm in Denver, Colorado and Summit County, Colorado sometimes. I'm here for those messages with something real.

Yeah, obviously treat others as you would like to be treated yourself, right? Love the process of entrepreneurship. We're lucky that we have so much of that in the US. Lucky that so many people are trying these things. And by definition, they change. So don't be too wedded to 2014 playbook.