What 20 Years of Software Investing Says About AI | Matt Hedberg

0:00I don't think optionality is free. I think every committed dollar, every time you say, I want to build a model agnostic product, every decision that you make, you're deferring something else. And I think these carry like real costs that don't show up even in like a P &L. This cycle is arguably one of the bigger, if not the biggest cycle that we've seen. Cloud really changed where software runs, where AI is really kind of changing what software does. What we're seeing is now this handoff almost back to technical leaders. Take Snowflake, for example, like when they brought Shridhar in, very technical focus, taking over from Frank Slootman, and it's put them on an AI scale that I think has changed the trajectory of the company.

0:41Five years from now, I think, the best management teams will have treated AI not as a product feature, not as a marketing narrative, not as a cost optimization exercise, but a fundamental re-architecting of how they create and deliver value to the customers. Like, I think that is the answer. Is this thing on? Yesterday's price is not today's price.

1:15Welcome back to Run the Numbers. My name is CJ. I'm a recovering tech CFO, and I am so pumped to bring you a conversation with one of my longtime friends, Matt Hedberg. For those of you who haven't read his work online. Matt is a sell side equity analyst at RBC. And perhaps no one on the sell side has taught me as much as Matt has. I've known him back since I was running investor relations at a fast growing cybersecurity company. And I know a lot of you out there right now feel the pressure to tell an AI first story because the market just expects one. But most successful businesses weren't built around AI.

1:47They are adapting to it. So me and Matt go deep on how management team should approach this communication gap, how they should talk about AI, whether they have it or not, and what competitive moats companies who win will grasp onto. All this and much, much more coming up. Matt, welcome to the Run the Numbers pod. Longtime friends, and it's good to be on the pod. I've been following your coverage for so long, probably coming up on at least eight years. I've been reading your stuff, listening to you ask questions on earnings calls. This is full circle to finally have you on my audio production. I've been a fan of your pod, of your blogs.

2:24The insight that you guys have is just fantastic. So the honor is mine. Well, before we jump in, for listeners who may not know of your work yet, can you give us a 30-second overview of what you do at RBC and what a software analyst does day to day? I've been covering software at RBC for 20 years, actually a little over 20 years, and I feel like a dinosaur. But sometimes that provides a little bit of perspective that folks that haven't been doing it as long don't have. Before that, I was a computer programmer myself. Probably not a great programmer, otherwise I would have been forming companies, but I come from more of a technical background.

2:57And while at RBC, I've covered just about every software company out there at some point. Today, my focus is primarily on cybersecurity, software infrastructure, broadly names associated with data, AI, analytics. I do cover a number of vertical SaaS names. For me, in more of the senior role. I mean, it's about information gathering and network building, CJ. I spent a lot of my time seeing companies, industry conferences, chatting with channel partners, just building networks. Obviously, there's an institutional investor side of it, but the life of a sell-side analyst too. There's a lot of stuff that happens behind the scenes that's maybe not as glamorous, like building models, writing unique research reports, being on earnings calls that you referenced on the topic of earnings.

3:40It seems like we're just on a constant earning cycle these days with both on and off quarter companies. It's a fun job. It changes rapidly, literally by the day, by the headline. And I just, I've been doing it a long time. I just find the day-to-day nuances just incredibly interesting. And later we're going to touch on that pace of information because I do feel like we're now living in this always on environment. You said you've been there for 20 years. Back when you started, it was probably like, hey, we'll pretty much hear from these companies four times a year. Now it's you're hearing from them nonstop.

4:10Whether some of this stuff goes through their comms team or not. I was doing some sleuthing on the internet and I went in the way back machine and you've had this front row seat to multiple tech platform shifts. And while preparing, I found this interview somewhere on Yahoo Finance, actually. It was from 15 years ago. And as I was reading, I couldn't shake the feeling. You were talking about cloud computing at the time, but I could have substituted in the word AI and just, and it could have been published today. On the constant news cycle, what I found really important in my job is having a well-curated Twitter feed.

4:46You'll find a number of interesting nuggets there that probably don't pass the comms team that I find incredibly useful in my day-to-day job. But yeah, on the topic of the paradigm between cloud computing and AI, it is remarkably similar in a lot of things. And I think we're going to talk about some of the similarities and some of the differences here. And in AI, it's effectively just another cycle. I actually used to be a COBOL programmer, so I can actually say that I was sort of part of the mainframe cycle. But, you know, you're a COBOL programmer, because isn't that one of the languages that Claude finally added to its arsenal?

5:17Exactly. Clearly, my job back then doesn't exist today. It's a little bit of a unique reference. That's where I got my start. And, you know, when I saw, you know, the mainframe to, you know, to the client server, to mobile, to SaaS, you know, these are all cycles. This cycle is arguably one of the bigger, if not the biggest cycle that we've seen. cloud really changed where software runs, where AI is really kind of changing what software does. I mean, that's kind of the simple paradigm that I think, CJ. You know, when you think about cloud, it really expanded the TAM for software vendors. I mean, there were more buyers, there were easier deployments.

5:51Really, when we think about like the bull case for AI, it is more in that TAM expansion mode. I think, you know, really enabling new workflows, new use cases, you know, that weren't able to automate before. And I think what still remains to be seen is how this plays out at scale. I think there's like three or four similarities that to me resonate with that conversation 15 years ago. And I think there's some differences. I think both cloud and AI represented generational platform shifts. They touch just about every industry, CRM, ERP, HR, security, infrastructure. And to the same regard, AI is layering into all these same categories.

6:24And so there's a lot of similarities between that. I think secondarily, when we thought about traditional SaaS, that was still a per seat model. Cloud operated, obviously in more of a pay-per-go or like a consumption model, like a snowflake. And AI is following a similar trajectory. Many of these AI native products charge per token, per query, per outcome, and it's not a seat-based model anymore. And so in some regards, cloud and AI are more similar in that regard. I think about third, when in cloud, you saw a lot of incumbents, think about Oracle, IBM, really scrambled to re-architecture their platform.

6:58While cloud-native startups, a Salesforce, a Workday or a ServiceNow, it really took a lot lot of share early on. I think in this AI era, there's a lot of similarities playing out. You're seeing legacy software vendors really start to enable agents or co-pilots or AI features, while native AI startups are really trying to displace them entirely. And so in some regards, there's a lot of similar to that. And then finally, in the context of IT spend, there were some similarities on the upfront hardware licensing, moving to ongoing, more of like an op-ex spend. And AI is doing a lot of similarities there.

7:29Companies are shifting budget towards inferencing, costing, model fine-tuning, things like data pipeline we hear a lot about, rather than one-time purchases. So there's a lot of similarities there. I think about some key differences, but I'll pause there for a second, CJ. That totally resonates. And 15 years ago, you had this quote, it was, now with cloud computing, we're almost seeing a little bit of reversion to a mainframe-like environment where you're seeing more processing in the cloud rather than on our own desktop and users interface to the cloud through an internet browser. Some of that It just felt eerily like some of the shifts are seen now.

8:01If the UI starts to become abstracted and we talk about headless software, a harness layer, it feels very similar to me today where, you know, you've got a lot of computing now, AI inferencing happening at the edge of the network. And you still have, you know, maybe a smartphone or a dumb terminal interacting almost the way a mainframe used to. There are certainly similarities to what we've learned from history. The thing that I focus on, too, there's a lot of key differences that are happening right now in the AI era. And I think this is where it kind of gets interesting because I think history can get us only so far.

8:32But when we think about the AI era, some of the rules are changing. Speed of adoption, CJ. Like, you remember when this started? It was like back in, I don't know, 2005, 2006, sometime around that when AWS launched. How long did it take ChatGPT to get to 100 million users? It feels like that was just months, right? And you think about, you know, enterprises deploying co-pilots, that's happening very rapidly. And so that speed of adoption is certainly accelerated versus the cloud era. The value accrual is less clear, I think, to me today at the app layer. And what I mean by that is like we thought about like the cloud, like the SaaS layer, a lot of value accrued there.

9:09And there were moats, like you think about some workflow and data and switching costs. That was very evident at the SaaS app layer. I think there's a real debate amongst industry participants that I talked to about, like, where is this value accruing? Is it the model layer? Is it the infrastructure layer? Is it the application layer? that's like really, really interesting to see where this value is accruing. And to me, that feels very different than the cloud era. Also, like I think about like concentration of vendors, like, you know, we all know there's three big hyperscaler vendors. And in some regards, there's a couple of big model providers.

9:40But I think what's different this cycle is there were hundreds, if not thousands of SaaS vendors at the top of the stack. I think it's debated right now, like how many application vendors will sit on top of that AI stack. It may not be hundreds and thousands. It might be fewer than that. And then the last thing, data and workflow moats. One of the things that I talk about all the time with people is moats. Moats change. And it's clear to me that the moats are evolving. When we thought about SaaS and cloud, the system of record, that was the moat. I used to be involved in migrating off of legacy systems to PeopleSoft back in the day.

10:11And now migrating off of Salesforce or Workday is painful. When we think about this AI era, really the model is commoditized. And when we think about the app layer, that could just be like a thin wrapper around data. So in some regards, switching costs could be lower with where we're going. And so really, we think about moats shifting. And that's, I think, a really interesting dynamic for where we're going here. Hey, thanks for listening. We'll be right back after a word from our sponsors. Maximoire, get your Maximoire here. Sorry, I just came from selling Cracker Jacks at the ballpark. Everyone in finance is adopting AI right now.

10:46The problem? It's a dozen disconnected tools. You have one for revenue, one for AP, one for the close, and each one needs training and documentation. And the expertise to run them lives in your head, not the systems. MaxiMor takes the opposite approach. One autonomous finance platform for the whole operation. Order to cash, procure to pay, the close, cash management, and reporting. It all runs on a unified finance context that pulls from your ERP, billing, banks, even email and Slack. No rip and replace. Maximo runs on self-learning agents that learn how your team already works. No prompts to wire, no workflows to build, nothing to document.

11:25Every output is audit ready and when a call needs human judgment, it escalates for review. Because the agents optimize, not just execute, and your KPIs keep on improving. One PE-backed customer posted 98 % of transactions straight to their ERP. No error rate. The other 2 % go to a human before posting. You pay only for real outcomes, not seats. See it at Maximor.ai. That is M-A-X-I-M-O-R dot A-I. Every decision a company makes is inherently a financial decision. Folks, I've been trying to tell you that for 300 episodes. So it's wild that most finance teams have to spend a week on what already happened.

12:07While my favorite movie is Back to the Future, we'd be living in the past. Speaking of movies, here's a terrible plot. Tired accountant chases down a receipt from March and then polices a$20 Anthropics subscription, starring the accountant Ben Affleck. So stop living it. You don't have to play defense on what already happened if you get your system sorted. And Brex's intelligent finance platform clears the backward-looking work. policy enforced at the point of sale receipts matched instantly close short into days not weeks what's left is the forward-looking part where to deploy capital how fast you can move what next quarter actually needs because the defense is handled by brex it's time for finance to go play offense throw the ball deep thousands of companies including anthropic coinbase and doordash already run on Brax.

12:56Head to brax.com slash metrics to join them. That's brax.com slash metrics. Today's episode is brought to you by Anarok, the intelligent tax infrastructure for global commerce. The world be changing. Companies that took five years to go global are doing it in just one. E-invoicing mandates are landing in every major market, and your team is starting to hand work to agents instead of logging into another app. Anarok's platform runs every tax job, and it's the only tax engine with reconciliation built in. Whether you're selling into your first state or your 40th country, Mr. Warwad, it's all supported by the same platform.

13:33One engine, however you reach it. Agents included. Every tax dollar owed, collected, because Anrock gets the transactions right underneath. Tax stops being an expense and becomes a pass-through. Anrock, built for what's next. Learn more at anrock.com slash rtn. I want to follow that thread, And we'll hit on management teams in a second. We can park that for a moment here. But most of our audience is not building an AI-native startup. They happen to work at a SaaS company, and the chat GPT moment happened. And flash forward three years, here we are. It's a brave new world. Most people are helping established companies adapt.

14:11What are the moats that are left? I think one of the most profound questions that I asked management teams, and this can be a management team of a software company. It can be a retail company. It can be a med tech company. It can be really any company. How do you think moats were yesterday? And how are these moats going to be looking tomorrow? And I think some management teams will kind of be deer in the headlights. They won't really have an answer or they'll revert back to yesterday's moats. And it's the insightful management teams that can actually perceive where these moats could be going. And that I find fascinating.

14:40And I've heard a couple anecdotal evidences of new moats. And I'd say what the foundational piece to it, like we've always thought data is a moat. And I still believe that. I think contextual data still is a moat. But the question that I ask is, what are you doing with that data? Because, you know, in a headless software layer, data is still important, but it's just an API transaction. And if there's no UI, like what's the value of that system of record? What's the value of that data? But I think some of these profound management teams are really figuring out new ways to harness that data. And I'll give you a couple examples.

15:11One management team that I talked to, they take all customer interactions, CRM, emails, tweets, whatever it might be. and they dump it into a model and they prompt it in real time. And they're trying to find out where are my customers pain points right now? They take that, they can feed it into a code generator and sometimes real time generate a fix that historically that management team might never know existed. But I think for these management teams that can act at the speed of AI and uncover these nuances in near real time, that's a competitive advantage. And I think the same could be said for whether it's customer service, broad-based product development.

15:46I think these management teams that can take data and integrate AI to it and really keep up at the pace of AI, that's a competitive moat that I think is differentiated. I mean, I think things like partnerships are obviously important. I think cash is important as a moat, right? Especially if you think of big incumbent vendors. These moats that we're talking about today, they're gonna continue to evolve. And I think betting on management teams that can see where the puck is going, if you will, those are the management teams that I wanna bet on. And I think it oftentimes start with great technical leadership.

16:17I've heard this framework that you have, that great management teams, strong platforms and large expanding markets are kind of the trio that you would want. How has AI changed that framework? For me, anyway, it's put far more emphasis on leadership. You know, when I think of some of the most enduring companies that I've covered, the ones that I would own for five to 10 years, leadership, platform, TAM expansion opportunity. But take an example of like a startup. Obviously, it's generally technically led. They build a great product that's disruptive and it's scaled to a certain level. And then at some point, they turn it over to a professional management team.

16:53It's often sales-led to try to drive increased adoption of this great new technology. But what we're seeing is now this handoff almost back to technical leaders. Take Snowflake, for example. When they brought Shridhar in, very technical focus, taking over from Frank Slootman, it really changed the trajectory, I think, of the product cadence of that company. and it's put them on an AI scale that I think has changed the trajectory of the company. Going forward, I'm looking for some of these management teams that can be very technical oriented. Gets back to the following, seeing where the puck is going.

17:24This era of sales-led executives, it's not gone, but I think there's an increased importance on the type of management team that you're getting. And that's not just the CEO. I think the CFO also can be very technically oriented, can look at customer problems and try to solve them. TAMs and moats and things like that are important. I think it all starts with technical leadership. But I also think too, that really examining these TAMs are important for where we're going. And I guess what I think about that is like this whole headless software experiment that everybody's talking about. Like I talked to leaders who interact with Salesforce data every day.

17:56They're not necessarily going in through Salesforce. They might be going in through a harness. Well, at that point, you know, you could argue that like that CRM data or that foundational data is still important. And maybe you're hitting that data, the system of record even more frequently. Maybe the cost per interaction goes down. But how do these companies think about expanding that TAM, getting more data, finding new use cases when maybe that UI is not the differentiator? So that's where I think technical leadership plus a market that feels like it can get bigger all coincides. And I think that's kind of how I think about finding really, really defining companies, especially when we're looking over like multiple decades.

18:31Maybe we could drill into what advantages you think incumbents may underestimate that actually become more valuable in the AI era, because you've brought up this concept of headless software. And just to the benefit of listeners, my layman's terms understanding of this is you're no longer interacting with the UI to get the data or whatever resources you need out of the system. it becomes more of a system of action where you're interfacing with it through APIs in the background rather than the traditional, let me log into Salesforce and check what's going on in my pipeline today. Because one of the questions that we get, CJ, is like, what does it mean for these legacy incumbent vendors?

19:06And it kind of gets to data still matters. But if you're abstracting that UI away from XYZ SaaS vendor, then we're like, what's the hook? Is there as much mind share if the person isn't living in that system all day? A SaaS vendor will be like, well, we've got all this cloud data and the customer is like, well, wait a minute, this is like my data. If I'm just like interacting it through an MCP protocol or an API, like why does it need to sit in XYZ SaaS system of record if I'm not even, you know, using the UI anymore? And so your question is like, how can incumbents win then in that scenario? From a practical perspective, I mean, I think there still are advantages that incumbents have.

19:42Proprietary data at scale, I think still matters. Like a company like ServiceNow has billions of IT incident records and resolution path and things like that. Now, that data is not reproducible. A lot of companies talk about sampling data or synthetic data, but like, I mean, that's billions of rows of data that in the AI era can tell you a lot of things about how a particular IT problem is behaving. And so while the model itself could increasingly be commoditized, I think the frontier models need access to some of this proprietary data. And so I think incumbents have to recognize that, but they also have to recognize that that toll bridge may not exist forever if it's my customer data, right?

20:22So then I think about what else do incumbents have? Well, they've got trust, they've got procurement relationships, and I think deeply understanding those is important. I mean, in some instances, as a bank, we've got decades-long relationship with some of these incumbent vendors, and that matters. How do you, as an incumbent, take advantage of that? How do you use that to your advantage over an incumbent, I think is important. I think partnerships matter. Like you're seeing a lot of my software companies create partnerships with these frontier labs. And I think, A, it's part of what's kind of killing the AI is killing software narrative.

20:54And it's giving a lifeline to some of these software companies that even a couple of months ago, the market didn't think existed. But I think these partnerships are critical. Even the notion of partnering with a frontier model, I think gives some credibility to some of these incumbent vendors. And I think cash obviously matters, right? I think, you know, using cash and your balance sheet is a strategic weapon. We went through this era, CJ, you know better than I do, of like, you know, growth at all costs to all of a sudden margins matter more. And you saw all these companies rush to drive higher margins.

21:25And really, while that was that sugar rush felt good for a while, I think, you know, there was almost a chance of over-indexing that, you know, almost took the eye off product product investment and R &D and even, you know, strategic acquisition. So using cash and the balance sheet as a strategic weapon, having an executive team that can identify an early AI startup, you might have to overspend for that, but doing so, taking out a competitor with your balance sheet, I think that's, you know, incumbents have that. And if you can use AI to speed up like your product development, like I said before, or just become, you know, more of a partner to your customer.

22:00These are things that are hard to replicate even for a startup. And so, you know, don't count the incumbents out. But I certainly think like we talked about before, you know, the game is changing, the rules are changing. And that's where I think management is incredibly important. Trusting that management understands this is important. Hey, thanks for listening. We'll be right back after a word from our sponsors. Remember when pricing was simple? One product, one subscription, one invoice every month. Those days are over. The AI economy is changing how everything gets bought and sold. We're seeing usage-based pricing, hybrid contracts, and something called a credit.

22:36The finance systems built for vanilla subscriptions weren't designed for this. This isn't just a rev rec problem anymore because outdated revenue architecture can kill a great pricing idea before it ever reaches the market. RightRev lets you recognize revenue in whatever shape it comes in. Whether you're launching consumption credits or testing entirely new commercial models, RightRev gives finance the flexibility to support whatever pricing model comes next. So product teams can keep on innovating and finance teams can keep pace. If you're architecting revenue for the AI economy, learn how RightRev can help.

23:09Visit rightrev.com slash CJ. That's me. That's rightrev.com slash CJ. Hey, founders and finance folks, you know how early on everything is fast and scrappy and then suddenly equity gets messy? Spreadsheets break, grant docs pile up, lawyers are sending new forms and collecting fees. It all shows up right about when you're already swamped. Well, if managing your cap table feels like one more frustrating thing on a very long to-do list, you need to know about Pulley. Pulley makes equity management simple and stress-free. You can issue options, model dilution, complete 409As, and more. All in one place with support from real experts when you need it.

23:49So if you're raising, hiring, or scaling, Pulley keeps your cap table clean so you can stay focused on building. not wrangling spreadsheets. Learn more, request a demo at pulley.com slash mostlymetrics. That is P-U-L-L-E-Y dot com slash mostlymetrics. I got news for you. The ERP category is finally getting disrupted. And if you haven't heard of Rilla yet, please pay attention. It's the AI native ERP built specifically to replace NetSuite. And it's already won over hundreds of finance teams. Their mission is to make the zero day close a reality. And they're actually doing it. We're talking teams closing the books at 1.35 p.m.

24:26on the first day of the month. Companies like Windsurf, Mercore, and hundreds of others run their entire finance stack on really. Revenue recognition, close management, multi-entity, native Stripe, and Salesforce integrations. Woo! Everything a scaling company needs. They've got 5.0 stars on G2. They're backed by A16Z and Sequoia. Heard of them and CPA-led implementations that get you live in 45 days. That is simply unheard of in the ERP space. If your books aren't running as fast as your business, check out Rillet. Book a demo at rillet.com slash CJ. That is R-I-L-L-E-T dot com slash CJ. That's me.

25:06We have a lot of finance people who nerd out over procurement. You said that procurement relationships can actually be a lever to pull on. I had a fascinating conversation with a CFO a couple of weeks ago. It all revolved around procurement. And what was fascinating to me is this particular company had years and years and years of data, of procurement data, like how contracts have been negotiated, you know, what this one customer says, what this one individual says. And this was all recorded. They dumped all that into a model and they effectively built a procurement agent where it was able to learn based on prior customer interactions and build a genie that, you know, the average salesperson or procurement team can use to help in contract negotiation at a customer level or a vendor level.

25:52In this era of, you know, everything's got to happen quickly, rapidly, you know, time to value is there. Customers are questioning the ROI, the level of spend. You know, if you can think about procurement in an AI era using this historical data and context, I think it's going to create, you know, an AI enabled procurement desk. And by the way, I think you're seeing AI native CFO offices being built where CFOs and finance team, FP &A teams use AI as a competitive weapon to differentiate themselves. And I think that's happening up and down organizations today. And that's, you know, non-AI natives, banks, insurance companies, retails becoming AI native.

26:28It's things like this. You're changing the narrative and you're able to innovate at the speed of AI. And that to me is a significant competitive advantage. The other pillar you mentioned was customer relationships. And a lot of people hear that and they say, well, isn't AI causing a lot of churn? Aren't we thinking about like just guarding the revenue base? How are existing customer relationships or some of these incumbents becoming a weapon? The scorecard is, you know, customers. It's retention. I start with gross retention. I look at NRR as a key metric. And I think if you've got a boat that's not leaking, you've got high gross retention, obviously you've got a happy customer base.

27:01Then you bring that customer along the journey. Can you convince them that what you're doing is adding more value and that should show up in NRR? It should fundamentally drive, you know, theoretically a reacceleration in growth right now. And so those customer relationships, feeling like you're being heard. Like I've had a number of conversations with CROs where even just like simple blocking and tackling isn't happening, where customers don't feel like they're being heard. They don't feel like their voice matters. In this era, theoretically, of lower switching costs, that could mean the difference between growth and decline.

27:33Any organization that I talk to can understand their customer better and understand those pain points and then listening to them. Because I think some of the best innovative software companies that I follow, they get their best product ideas by listening to the customers and understanding how they can help them, how they can add value, how they can solve pain points, getting back to the moat. I think if you can do that effectively, I mean, that's a moat in and of itself, but incredibly important. I think of an example of Microsoft Copilot, even GitHub. And you think about how could they possibly ever fall behind or let someone like a cursor come up with an install base that's this big.

28:07And so I think it's the age old adage of can the incumbents innovate faster than innovators trying to find sales and commercial routes to market. There's probably a question in there somewhere about like, how do you think about protecting your legacy revenue base versus disrupting it? And yeah, I mean, I think, you know, in a lot of these companies, you protect the sacred cow, but there's risk to that. I think, you know, in some regards, you have to do that. These companies that build speedboat teams that really try to disrupt themselves. That's something that we're seeing a lot these days, especially from, I think, more of the innovative companies and anticipate change.

28:40Just go in Zoom, going back to the management team, seeing this and being technically oriented to be able to understand this stuff. Theoretically, you could be able to build, buy, or acquire that technology before it becomes a billion-dollar company. That's the challenge, but it's also the opportunity for these companies. I don't know why I just had this visceral reaction, but I'm remembering having that physical phone bridge in the center of a conference room before we used to even do the video calls. It was like GoToMeeting. I forgot what it was called, but I had to dial in, and I was like the youngest person in the room, so make him do it.

29:09And I remember fat fingering in the number twice in a row and having to start over, and everybody's waiting for the call to start. So I'm just so thankful now that we've moved to having links for meetings. Some of the stuff you take for granted. I mean, like I think about my career 20 years ago. I don't know how I like existed 20 years ago in a career. Our kids, when they get into the workforce, it's incredible to think of. It's going to create a better product, whether it's a conference call or a piece of software that's being output. But that's the excitement that I get in this sort of this very, very unique era that we're in.

29:35This difference between protecting your install base versus disrupting it. There are a couple of examples of companies who have managed to pull it off. Intercom, which became thin and underwent a huge acquisition by Salesforce, I believe. They burned the ships and they went after their own install base. That's a really brave thing to do. There's probably a sequence here that's important. You could probably burn your base too quickly and the street's not going to give you credit for that. But I think it's really like an art more than a science here. And I think if you're a company and you've got this durable revenue stream that the paradigm isn't there yet quite to fully replace it, like you're in the early innings of something.

30:11I think Adobe and Microsoft saw this. Adobe didn't flip the switch overnight. I think, you know, their on-premise business still generated enormous cash and it helped fund their move to the cloud, you know, whatever that was a decade ago. The install base really was an economic engine, I guess, if you will, that helped pay for the future. When we think about when protecting the base becomes a bit more of a liability, that's when it becomes a little bit dangerous. And I think I've seen this where the company is no longer able to really choose to protect the base, but they actually are structurally unable able to cannibalize it.

30:41And that's when I think you've waited too long. This is a really hard problem, I think, for specifically incumbent software vendors to think about. When I think about this, I talk to a lot of executives and the framework that I use is, rather than kind of like a binary, like do you protect it or do you cannibalize the base? I think the best companies I've observed really manage this as a portfolio with a question like, what's the half-life of our existing revenue stream? How long will it take for the market to structurally move away from it, regardless of what we do. And if like, if that half-life is five years, you know, you probably protect and fund that.

31:14But if like the half-life of that base is like a year or two, you know, I think you cannibalize it aggressively because, you know, it's either you cannibalizing or the market's going to cannibalize it. And so the strategic sin here, I think is really treating, you know, shorter half-life revenue as if it were longer half-life revenue and even the inverse of that. Well, let's talk about capital allocation because CEOs, founders, and CFOs have to get in this room and think about how you make one of these transitions, because in many ways, what you're willing to cannibalize is the thing that is allowing you to have the runway to even go and do that.

31:45It's a thing that I think about, especially because I cover a lot of like, you know, more mature companies and like, how would I allocate spend for, you know, infrastructure or customer experience? If you have an AI leaning CFO office, I think they're going to be able to use a lot of data for insight. So when I think about like, you know, how would I fund these various things, you know, I think about like data infrastructure first. I don't think it's the most exciting element of funding, but I think it's the constraint that really kind of is the governor to everything else. Like if your data estate is fragmented, if there's no telemetry, any AI that you built on top of that is, it feels like it's destined to fail.

32:19I think, you know, the mature software company that can think about their data as an advantage and we're, you know, going back to the earlier comment, but like look at ServiceNow CMDB, like that's a foundational layer, if they are able to take that foundational infrastructure layer and ship AI features on top of it that actually work at enterprise scale, that should get traction. In my head, I'm like, start with data infrastructure first. Then I think about product second. And I think for AI native product capabilities, it's second nature for them. If a competitor starts to enable intelligence of the workflow and you don't, there's going to be instant pricing pressure on you.

32:54You're going to see churn, you're going to see pricing pressure, you're going to see just disappointed customers. The nuance here is like, you know, if product AI development is done properly on a solid infrastructure layer, then it's really a pricing and packaging opportunity. And I think that AI layer, you know, helps companies move from selling seats to selling outcomes. And so I think like that's part of this product maturation, internal productivity. To me, that's kind of the third thing. I think if you get your data architecture right, your products right, If you can then say like, how can I use AI internally, co-pilots, automated QA, especially for some of these bigger companies with like thousands of employees with well-defined processes, the payback there can be substantial.

33:33But if like, if you're doing that first, you're sort of doing it without a real purpose. So to me, like that would kind of come third. And then finally, this is maybe a controversial take. Thinking about the customer experience almost last in that, not because the UI doesn't matter, But like, if you don't do those first things, I don't think, you know, any kind of user experience is going to really matter, especially like we talked about that headless experience. And so if you nail AI in the product, the CX benefits will follow. And in other words, standalone CX benefits, like a chatbot on bad data, that's not going to do anything.

34:04So I really think there's a sequencing here. And I think as a CFO, there might be a temptation to kind of spread these investments thinly across maybe all four of those categories. And to me, that's a mistake. I think you need to kind of sequence these things properly. And I think it's going to take time, but I think that's the right outcome. If I'm a CFO thinking about strategically deploying capital. I love that. And I really appreciate how you prioritize them because some of these make the others better if you do one first. For sure. You know, maybe the natural knee jerk reaction is let's just use AI internally, you know, as aggressively can.

34:36And then like, you know, that works for a while, but that's like a sugar rush that wears off. But I think if you can build a strong foundation on data, infrastructure, and then AI products, Like that is like how I would think about strategically allocating capital. Where do you consistently see companies overspending? I don't think we know. And from my like outside in C, I probably have less visibility than if I was inside the company seeing how things are going. For me, like conceptually, you know, and I used to be a COBOL builder. So, you know, I do have a little bit of and by the way, I did more than COBOL.

35:05I did a lot of C++ and other stuff that's a little more exciting than COBOL. But all this money being spent on just building proof of concepts without a clear path to production and even more importantly, monetization, there's some overspending going there. And so if you look at a team and they're spinning up a bunch of like AI pilots across all these different departments, you've got a lot of fragmentation. I think six months later, one or two of those might show promise. But like if you think about the organizational cost of running all of these, you know, there's like vendor constraint issues.

35:33You've got dedicated engineers. Some of these bad practices can kind of compound quickly. So really the failure isn't just that like the individual pilots don't work. It's that there's like no triage mechanism to kill ones that don't scale and move quickly through that and really identify, you know, kind of that killer app. It's like the shotgun approach. I think you've got to be really, really smart with finite capital. And then you've got to have a path toward taking these proof of concepts to production and then ultimately demonetization. And I'll tell you, CJ, like I've talked to a number of software companies who've got in their minds and candidly a really strong AI product.

36:05The question that I get from customers is like, it's just like utility at that point. Having a good AI product is like, that's going to save you from churning. What is it that's going to actually cause me to pay you more money as a vendor? And that's when I think like that vision to focus on a few strategic bets and get that all the way to monetization. Like that's what investors are looking for. They just don't want hand waving about, you know, random AI metrics that have no bearing on like revenue at the end of the day. So, yeah, I think that's probably one of the biggest sort of like wrongs at this point from from like a capital allocation perspective.

36:39We talk about one way versus two way doors when it comes to decision making on the podcast a lot. Matt, how much optionality should companies preserve here? Is this a moment to make one big strategic bet? I don't think optionality is free. I think every committed dollar, every time you say I want to build a model agnostic product, every decision that you make, you're deferring something else. And I think these carry like real costs that don't show up even in like a P &L. You know, they show up, I think, in slower time to market, engineering complexity. And I think the companies that have really pulled away in my mind over the last 18 months or so in this AI era, you know, they've made opinionated bets early on how they can truly leverage AI from a foundational layer.

37:21And they've integrated into their workflow and they've really illustrated to customers that there's true value that I'm delivering. They didn't wait for necessarily maybe the perfect information. They picked a direction. They shipped a product. They think about monetization. And I think that optionality is what matters because I think the worst outcome isn't necessarily making the wrong bet. It's like making no bets at all. And then you sit here 18 months from now and you've got a bunch of half-baked features with no clear narrative. Customers at that point, it's just easier to choose a competitor.

37:49So yeah, that's kind of how I think about it. I think, you know, optionality isn't free, but you've got to be bold. And I think you got to be very strategic in these bets. And that, I think you're seeing some of these companies and look at some of the EV to sales multiples of some of the highest value software companies. I think that's where they've succeeded. I think in this era, it's never been more important to have an opinion. The cost of not having an opinion, you kind of become a dinosaur pretty quickly in that era. I think it compounds, actually, the faster things move. I couldn't agree more.

38:14I see that day to day, whether I'm talking to companies, whether I'm questioning my own opinions. I think that matters more. Well, one of the hardest parts of a platform shift is separating the leading indicators from the lagging indicators. What metrics tell you a company is actually building a durable advantage? The single most important metric in my mind, and I would say gross retention is obviously important, but I think it's expansion velocity within your base. And this just isn't, it's not net retention. And we all get NRR from these companies because that's a lagging indicator. I think specifically it's whether these customers who are activating the AI capabilities are actually expanding faster.

38:52Is there better retention? Do they move up market to SKUs faster? And I think if an AI feature is generally solving a problem for a customer, think about it as almost like a wedge that pulls user even deeper into the platform. So when management teams, in my mind, if they can't quantify kind of cohort level expansion, I think that's a problem. And when they can be more specific on that, to me, that's a great indicator of success. Lots of companies that I talk to, they'll say, you know, we've got, you know, I don't know, X thousands of customers who have activated our AI feature. And in some regards, I think about that as like, well, that's great.

39:24I don't know that that tells me anything. It's like window dressing. But I think, you know, if you can talk to a company and they can talk about like, like we're seeing durable signals of daily or weekly active usage per seat, is that user engagement growing over time? Like that to me is a real defining feature of you've hit the mark, you know, a feature that compounds in usage. Like that's where I see the real value. And that's where I think you're becoming part of that workflow. And that's where it becomes less of a novelty and more of a habit. You've got like a data network effect. Is that actually showing up from a customer expansion perspective?

40:00And like, I'll take an example that I dealt with years ago. Like CrowdStrike has this thing called a threat graph. And I think that's a perfect example of like a company building this incredible cloud data emote. And every endpoint that they add makes the detection better for the next one. And what they do with that data is they're able to spin up and build new products. Sometimes before the customer even recognizes they need the product, that's a real judge of like future success. Maybe the last thing I think about is like, you know, when I think about like attach rates, I think about, okay, like what about like on new bookings versus, you know, new customers?

40:34And I think there's an important distinction here, like high attach rates on new deals. Like that tells me that The market is pulling in your capabilities, your AI capabilities, wherever your new features are. You know, customers are demanding it as part of that initial purchase decision. I think, you know, if you've got high penetration of your base, you're successful then in upselling your customers. That's important too. But I think looking at new product attached on new bookings, that to me is a really insightful conversation that management teams can articulate. You know, I think there's a bunch of other things.

41:05I could skip through some of these things, but I think like, you know, it's easy to, you know, for a lot of people to want to go to a frontier lab. But if you're not attracting and specifically retaining the best talent, that is troublesome. That's a sign of, you know, future issues at that company. I love net dollar retention as much as the next person, but I really dig how you pointed out the expansion velocity is so important because when we get served up these NDR numbers, Matt, they're usually on a trailing 12-month basis. And it's very hard to move the needle in the near term if you're looking back that far.

41:36For sure. And I think that's the kind of the false narrative on that metric. And so I think to the extent that we can turn that into a more of a forward-looking metric, one that could project the future, those are some of the signs that we look for a change in velocity of a company. We don't have the perfect metrics being outside in, but I think companies, management teams that can project that to investors, that's what I'm looking for. There's so many games that can be there. Like, were they forced to put it on? Did they get it for free? Did they have to pay you for it? And that's why I think the land of new customers who are pulling a new product forward is so telling because they didn't have a history with you, right?

42:11We've been hit over the head with like land and expand. If you're truly building something differentiated, you should be in a position to land even bigger, I think. Now, maybe that's, you know, there's probably a trade-off between a little longer sales cycle and things like that, that I certainly acknowledge that. But I truly think like when customers want the product, you're being pulled in, it's driving true ROI. That's the distinction of having a great AI feature that's just a utility and it's going to prevent churn between a company that actually has, you know, real expansion opportunity. And the customers are like, I'm going to pay for that because I can't get that anywhere else.

42:44What role does a great CFO play during a platform shift? I think it's a really dynamic time to be a CFO. Like one CFO that I talked to said he's trying to build an AI native CFO office. And I'm like, that's an incredibly interesting concept because I think there is no playbook for that. Like that is a new concept. You know, using AI as a weapon, even in the CFO's office, I think really, really matters. And so like, that's one of the things I'm really curious on how that playbook unfolds. And I think we'll see some CFO offices emerge as thought leaders here. And it could very well come out of some of the frontier labs.

43:16But I think, you know, having a strategic CFO, somebody that's a little bit almost like a COO, I think is very, very important. And I would say, too, a CFO that can talk the technical lingo, I think is just as important because at the end of the day, I think they've got to be able to understand what the customer wants, be able to sanity check what the sales forecasts are telling them based on product knowledge and product acceptance. And so I think the era of a more dynamic CFO is upon us and ones that can be more AI leaning. I think that's that's super important. I think the best CFOs, I think, during platform shifts, and I'm thinking about like a legacy lens here, I think about them as capital allocators, not just accountants.

43:53And we kind of touched about this a little bit, CJ, but I think, you know, that distinction probably matters even more. And you asked about like the proper ways to think about allocating capital. I think that's, you know, an accounting minded CFO, I think is going to optimize for margins, stability, predictability, you know, guidance beats, you know, investor comfort. And listen, that's all important stuff. But I think during this AI transition, the job function of a CFO really about deploying strategic capital from legacy cash flows towards emerging opportunities at the right pace, fast enough to not miss a window, but disciplined enough to not destroy the balance sheet and credibility.

44:29I think that's one where a CFO treats the P &L really like a portfolio. And you're actively managing the harvest versus invest and really specific timelines around deploying and then killing certain investments. So, yeah, I really think that these dynamic CFO offices are going to emerge. And I think they're going to be very, very AI leaning and very strategic in how they're deploying capital. Up next, it's time to ask some better questions. And you know who asks the best questions? Our friends at EY. Across enterprise transformation, risk, audit, diligence, your advisors at EY. Let's ask some better questions.

45:08Gearing towards a close, I want to talk about effective communication because I feel like companies are having to relearn how to communicate their AI strategy. There are all these expectations around having one. This is very meta, Matt. Learning how to talk about it and get that point across, that's also changed. Where I've seen companies fail in my job is they might have the best solution, they might have the best product, but if they can't cleanly articulate it, clearly to investors, then I don't know how it's going to resonate with customers. And so I think for me, when I'm listening to some of these C-suites, for me, I think the single biggest tell that I'm looking for right now is, do they start with the customer problem or do they start with the technology?

45:49And I think, you know, a CEO that describes their AI strategy by saying like something like, you know, our customers have always struggled with, you know, XYZ and now they can finally solve it because of ABC. Like that's a company where AI is being pulled into existing value proposition. And the contrast of that is when they start with something like, like we've built a large language model that can do this, or we've integrated generative AI across our platform. Like that's a company that's pushing technology into the market and they hope that demand follows. The first thing when I think about this is really the framing of it suggests that the company understands why customers are paying them.

46:24And it suggests that they're not going to be left behind. So to me, do they start with the customer problem or do they start with the technology? That's a tell to me. The other thing that I think about is like incumbents that I trust the most, they can often draw a straight line between like their pre-AI strategy and their AI strategy without, I guess, it feeling like forced. The narrative really shouldn't require, I think, a sharp pivot because, you know, some of these major teams that we talked to like, you know, we've been doing AI before it was cool or before it was a thing. And like, you know, take like Palo Alto, like in this era of like platformization, they've talked about AI and ML for years.

46:58And now they're using AI to kind of correlate signals across their security portfolio. To me, that's an extension of the thesis that they've been building for years. AI in that case, I think can make their strategy better. But the strategy existed even before Gen.ai and Open.ai existed. And you contrast that with companies that really had no coherent platform strategy recently, and suddenly they claimed to have AI. That's just hand-waving, I think. And I think that kind of retrofitting can sometimes make the problem worse. And so I think kind of thinking about a continuation of existing cycles, to me, that's something that I look for.

47:32And then when I think about, I spend a lot of time looking at like org structures now, because I really think how companies are building their own leadership is important. Like where are they putting AI in the organizational structure? I mean, I think it should be embedded in every employee from the CEO to the CFO to the head of sales. If AI is kind of just a separate innovation lab that reports to some AI officer and there's no P &L, I mean, I think there's a real disconnect between their strategy and the core problem. But if like AI capabilities are embedded within every role in this organization, And it often depends on who these people are, like the PMs, you know, the engineers, if they own the features, if they own the AI, like that tells me that they're treating AI as an integral part of their strategy, not just a bolt on.

48:15And so I spend a lot of time really just like looking at LinkedIn and just following, you know, where executives go. And like I mentioned this earlier, I think the inverse of that is when you're seeing a departure of some of this. To me, that's a sign of maybe some bigger issues here. And so those are some things that I kind of think, you know, the strategy and things that I look for, those are like super, super important to me. I really love how we've been talking about technology, but you drew it back to the people, right? That is truly the leading indicator. It's the people, it's the technology, and it's can you expand that TAM?

48:47We were all like beat over the head, you know, a couple months ago saying we're all going to be eliminated. And I think in reality, AI serves its purpose. You know, agents serve its purpose. And when I talk to, you know, executives, I talk to investors, when I talk to my kids, you know, I don't necessarily worry about AI displacing me. I think like if employees themselves can enable themselves with AI, that's the disruption. And it starts with people. Okay. So me and you are hopefully recording five years from now, right? Hopefully our wonderful sponsors keep letting me do this for a living. You've watched multiple tech revolutions over two decades here.

49:20When we look back at this AI transition, what do you think the best management teams will have done differently from everyone else? The answer might be surprisingly simple. Like, I think we might look back and be like the Mag7 may alter a little bit or, you know, there's obviously going to be new AI startups that emerge. But five years from now, I think, you know, the best management teams will have treated AI not as a product feature, not as a marketing narrative, not as a cost optimization exercise, but a fundamental re-architecting of how they create and deliver value to the customers. Like, I think that is the answer.

49:55And there's a number of companies today that I think have already recognized that. This gets back to the optionality question that we talked about before. I think we're going to look back and some of the best management teams, they'll have made fewer bets, but they've made bigger bets and they've made them early. They're decisive in what they're doing and they're right. And again, easier said than done. But I think, you know, looking back at it, there's a number of examples. History will tell you that that's played out every cycle. And I think this cycle will be very, very similar to that. And, you know, I think, you know, some of those conversations could be uncomfortable when we're thinking about, you know, the future, but that's like a core tenet of what we think.

50:30It gets back to like, I think success is going to be natively integrating AI into every employee experience. Maybe the last one, and we've kind of talked about this before, is just anticipating moats. I find, again, I'll kind of go back to where we started. Those management teams that have at least a view and a vision of where these moats are evolving, where their prior moats are dissolving, that's the kind of company that I want to bet on. And those are the ones that I think five years from now, we're going to be like, that's a big company. and it was very clear of like how they got there and, you know, how they, you know, why they didn't fail.

51:02Matt, this has been an absolute blast. You're such a great thinker and communicator. I really appreciate you coming on the podcast. What is the old saying? First time caller, long time listener? That means the world is coming from you. Yeah, well, yeah, I appreciate your insights here. We've known each other for a long time, so it's fun to spend a little bit of time talking about some technology and AI trends. Absolutely. Run the Numbers is a mostly media production. Yelling an intro by Fat Joe. Artwork by Meg D 'Alessandro Show is executive produced by Ben Hillman Nothing said on this podcast is intended to be business or investment advice It's the sole opinion of me A guy who feeds his dog way too much ice cream And has a history of net operating losses LOL If you like this podcast, hit subscribe and give us five stars It will take like two seconds And our algorithm overlords love it Drink water, call your mom and have a great day Peace