Loading summary
Transcript32 lines
- [00:00]
A
Foreign. Welcome to the AI to ROI podcast, the big Story episode. I am Ray Reich, the founder and.
- [00:16]
B
CEO of BenchmarkIT and I'm Peter Buchanan, the managing partner at New Plan.
- [00:21]
A
In this weekly Big Story episode is where we dive in to that big story of the week from the AI to ROI newsletter. We which can be seen and read in detail on substack. That is AI to roi. That's the number two. So AI to roi.substack.com so Peter, can you introduce this week's big story that we'll be discussing?
- [00:46]
B
I can. So first of all, welcome to the AI to ROI story of the week. This week we're going to be talking about a huge issue for software companies. Its transformation of SaaS companies to become AI first companies. It's a huge challenge and there's a lot on the line because the SaaS industry itself generates $273 billion a year in revenue and it is under attack. So if you listen to Salesforce founder and CEO Marc Benioff, he says we are rebuilding every one of our products to be agentic AI. So Ray, let's get started here. This feels different than past platform transitions. And so why is that? How is this different?
- [01:39]
A
Well, first of all, even though we do not provide investment Advice here on AI to ROI, if you look at the enterprise value of SaaS companies, the enterprise value to revenue multiples, I mean in the last six months, ServiceNow down 22%, Salesforce down 45%, multiples down to about 5x at median. So there definitely is, the investors are saying there's a transition. We see all the AI hype now the AI hype is more today around hardware infrastructure, you know, than the nvidias the I believe at Google with Gemini 3 OpenAI. But you asked me the question, how's it different? Well, I think there's some similarities Peter, to the transition we went through about 20 years ago from the traditional on prem enterprise software model to software as a service. Now there were phases of that. I still remember the first phase and this was kind of late 90s, early 2000 was something called ASPs, Application Service Providers. And this is where enterprise software companies like Siebel and Oracle or weren't ready to move to the cloud or away from that one time big upfront license because that drove a lot of their revenue growth and their multiples. So these ASPs were created and they hosted enterprise software inside of their own data centers. It wasn't multi tenant, it was one server, one customer. So they put an instance of SAP on server one for customer one and server two for customer to. Now one of the major barriers of ESPS is they were beholden to the actual oem. SAP could dictate what they could sell, how they could sell it, et cetera. And then there were a few brave enterprise software companies who early on said, hey, we believe, number one, the subscription model is the future of enterprise software. And number two, we do believe that multi tenant architecture is, is a great efficient way to deploy because it removes some of the total cost of ownership from our customers. So a few brave companies such as Splunk, MongoDB, OneStream, they aggressively moved from an upfront enterprise license to subscription. And then there were VCs. And I still remember this because we were bought by Vista Equity when I was running sales services and marketing for a enterprise software company. And part of their model, their financial engineering model was eliminate selling one time license, move everything to an upfront license, even though it was still on PREM software. And for existing customers they started charging them an annual subscription, basically replacing their maintenance fee. So I guess what I'm saying is I don't think it's that different to go from SaaS to AI. And I think the on prem software to SaaS model provides us some interesting insights and lessons learned.
- [04:55]
B
So is this just about adding AI features or is it about rebuilding the SaaS company from the ground up to be AI driven, you could have asked.
- [05:05]
A
The same thing about trying to turn your enterprise software company into a SaaS model 20 plus years ago. It's much bigger and it's much bigger than putting a thin user experience fail like a prompt engine on top of a legacy enterprise application, a SaaS app, every function needs to be fundamentally transition to think AI first. This means marketing needs to think differently about how they're doing. Demand gen pricing has to be totally redone. Sales has to think about how am I doing both my targeted outbound demand gen and how am I managing inbound leads finance. We're going to talk about this later I think Peter, we need to talk about how budgeting is different. COGS and CAC have some different profiles in the AI model. R& D. Hey, we need to start thinking about not only automating some of our coding, but thinking about things like context graphs and agentic AI. So it's not a slight transition. It's a fundamental rewrite of an operating culture.
- [06:09]
B
We had our old playbook. We had 20 years of predictable margin, strong retention seat based pricing. Software vendors are used to it, customers are used to it. Now AI native companies come along and they're just literally playing by different rules. Now you've got companies like lovable that hit 200 million in AR in 12 months. With under 20 employees. You could cursor hit $1 billion in under a year. And they're not just adding AI, they're selling something different. They're selling changing work inside these companies. So this seems like it's a two front war for SAS incumbents. The first is they have that internal refactoring that you're just talking about legacy architectures, pricing workforce. And you've got external disruption because you literally have new types of companies that are shiny doing things differently, using their cash much more efficiently. So how do SaaS companies get started fighting this AI native threat? What are the first three things that they really need to do?
- [07:21]
A
The first thing I'll say is the two factors you talked about. You're not going to change the external variables that are disrupting the software industry. I still remember people say, well Larry Ellison said in 2000, 2001 that SAS is never going to overtake on prem enterprise software model. But on this side he invested in both Salesforce. He was one of the earliest angels. He also invested in NetSuite which was going against their core financials. So even Larry knew at least he was hedging his bets. There's no hedging bets here. AI is going to be at the center of the software industry going forward. So first thing I would do, and by the way it's this is not rocket science. I've been working with a lot of legacy SaaS vendors over the last three years and they are doing this. So this is based upon what I see happening and what I'm coaching CEOs to do. You need to re enter re architect the entire company around customer outcomes. Now you say well wait a minute, what about AI infrastructure, et cetera? Well, the big promise of AI is selling work and selling outcomes, not selling software and automation. So you want to re architect your entire company around customer outcomes versus product feature functions. That means messaging, positioning, pricing, even the workflows that you're automating today need to be thought about in an AI first model. So you're re engineering every process to be AI first, whether that's your outbound and how you're doing personalization and how you're differentiating your outreach to ideal customer profile 1 versus ideal customer profile 2. Your whole inbound process. You're going to be able to much more dynamically use AI and automation to determine should that lead be given to a human? Should it be further nurtured? How do you score that lead? How do you personalize the outreach? How do you enrich the lead customer success? We're going to move from being more of an uber account management function to doing things like let's use AI to automate the onboarding, get higher product utilization, use in app recommendations based upon a user being using feature three. And you know that feature five is where there's higher value. So that's the second thing. And the third is, and this is something that I Learned from the CEO of a Fortune 50 company, whenever a leader, a senior leader was leaving, the first thing he asked the new hire to do was how can AI change every department and every role in your organization? Map out the existing org structure and initially the top roles and determine how AI can fundamentally re engineer the organization structure in the process. Simple examples, if you're in a SaaS company Rev Ops, how many of those should become go to market engineers? How do you take a professional service consultant and make them a forward deployed engineer that can use their work to bring right back into the product in a much shorter timeframe marketing operations, maybe you become an aiops person. Hey, how do I deploy agentic AI for my marketing campaigns to make a more human resource effective and more efficient? So those are some examples.
- [10:48]
B
Peter okay, so this is truly a heavy lift. The way we're talking about it so far it sounds kind of dire for SaaS companies, but I'm not sure that's so because I think SaaS companies have incumbent advantages that AI companies have to match. And I'll list six of them here and then we can go through and discuss how SaaS companies can make this disc go over to being AI driven while retaining those advantages. So the first one is SaaS applications are systems of record. They're critical. They're embedded in workflows throughout the company. Two, they host crown jewel data. They're basically systems of truth. Three, they're embedded in every workflow executed by every human for their points of integration. So data flows not only inside the SaaS applications, but between them and into data warehouses and out to customers and out to suppliers. As a data conduit, they're essential. These companies have sales teams, they have customer success, they have champions inside their customers, they have partner network, have brand trust and also they've passed the cybersecurity data privacy requirements to operate inside of a mid to large sized enterprise. And they have features built in that allow them to understand like context Graphs which are relatively new, but sort of catching on like crazy right now these applications have an understanding of what's actually going on inside the company. So when I look at these six traditional SaaS moats, my question to you, Ray, is how do the SaaS companies moving to be AI driven maintain and expand their moats against these aggressive AI native companies?
- [13:08]
A
Those are great differences that truly can be leveraged to not only ensure you have that moat, but to make that moat deeper, wider and less penetrable by these newbies. Same thing happened back in the early 2000s. You know, everyone said people aren't going to leave Siebel. All their customer data is already in the Siebel platform. You're not going to be moving off SAP. All your financial transaction history is an SAP. But people did move to Workday, people did move to Salesforce. So what do we learn from that? So being a system of record versus a system of action, a system of record is, you know, all your transactional data and entity data resides there. That's very different than if you're a sales enablement piece of software that's doing outbound cadence design. So in a system of record, you want to make sure you can really lock in that position. One of the things is look at the points of integration and try to continue to enhance those points of integration. Look at your current automation workflows that are human enabled and human assisted to be more AI agent executed. How can I take those automated workflows and turn them into agentic AI processes? Those are just a couple examples. This is not time for an exhaustive list. Data. Hey, if you've got all this customer data, you know your customers have been having to export that, normalize it, put it into a data house, a data lake. How can you build with AI at a decision intelligence layer on top of your transactional database? That's an example of how you can turn that data into an unfair advantage. Workflows, right? As I said earlier, workflows are not just they're initiated by a human, they require human data entry. Right? They push a button and then it progresses that process. You need to think about how an agent actually ingest something from another system or another person and it kicks off a workflow that's intelligent in design integrations. We talked about integrations being a great moat. Hey, don't say, hey, I've got 29 connectors using a REST API. Do you have an MCP, a model context protocol infrastructure? Have an aggressive new partner API program to get a lot of the New AI extension software, build some integrations to them, and then on the distribution, hey, one of the things I'm saying, if you're a system of record, you've got big SI partners. Sit down, understand your SI partners kind of where they're going with AI, building out AI infrastructure, et cetera, and make sure that you're part of their recommended solution for AI transformation. So those are some examples.
- [16:02]
B
PETER okay, so, and those are natural courses that customers will follow. So what do you think the motivations for CEOs are looking at this? What's driving them to make this transition to AI driven?
- [16:21]
A
Well, we covered this in one of our big stories a few weeks ago, the AI ROI newsletter. It's the new AI metric. It's months to 100 million. It's ARR per FTE and SaaS companies, man, we're really happy to get to about $400,000 per FTE. We're AI native companies are saying, hey, my minimum threshold is a million dollar per at per fte. So that's motivation in and of itself. You know, whenever there's a technology disruption that is, we're moving from one generation of platforms to another. A lot of companies are saying, hey, I want to invest where the puck's going, not where it's been. So if you're going to compete in the future, number one, just look at SaaS multiples on the public market and you know that you can't be there. And number three, look at valuations and where VCs are putting their money. You know, over 65% of VC investments in the last 12 to 24 months have been in AI native companies. So we know where the market's going. And if you want to go to where the market's going, you got to basically do a wholesale. We're changing to an AI first culture. So that's what I say.
- [17:35]
B
PETER okay, so let's break down four of the key functions, the four key functions inside of this SaaS company moving to AI driven, and talk about what has to happen in these functions to be successful. So the first one here is product development. So most SaaS companies, they're using new AI tools to improve their products faster. They're also using AI tools and go to market. So do you have a framework or perspective, beginning in product development, for what companies need to do to produce at what I would call the speed of AI? Given the pace of the market today, what needs to change there? And then connected to that, we'll just move right in after that to what has to change and go to market to get those new AI things that you're working on in the hands of customers.
- [18:35]
A
You know, there's all kinds of benchmarks out there, Peter. You know, I've seen reports that say that CTOs are reporting anywhere from 50 to 75% of code is now AI generated. I mean that's up 3 to 5x over the last 12 months. I mean even our CTO who's a 25 year veteran now is doing almost 100% of his bug fixes using cloud code. And about 50% of his new engineering development is being done that way. It significantly compresses software velocity. So notice I didn't say productivity, but velocity increases dramatically. But at the same time when you're trying to compete against AI Native companies, because AI native companies, they're spending 60 to 70% of their R&D on new AI feature function. Some of it is more less AI specific things like organization structures, permissions, security, et cetera. So you've got to be able to get a lot more of your R and D resource off your existing legacy SAS feature function maintenance, you have to do that. But code tools do that for you where you can take your engineers and architects and spend a lot more time there. So that's what I say on R and D. Peter.
- [19:49]
B
Okay, so go to market also huge changes there. Just the whole way salesforces are run and customers are one and onboarded. It's a huge change for companies. And if you don't change, you're really left behind. So how does that work?
- [20:11]
A
Well, where a lot of organizations have started is focused on lead generation. In AI native companies their costs per lead are anywhere from 40 to 60% lower because they're doing so much of it through AI agents and automation, sales cycles times because they're being much more targeted and engaging with their icp. And even for inbound leads, they use agents to nurture poor fit leads and people to close great fit leads. So you're going to see higher win rates, shorter sales cycles times. And most of these AI native companies, even in the early days when they're growing from 1 to 5 to 10 to 25 million, they're only consuming 20 to 30% of revenue on go to market resources where traditional SaaS companies would spend 40, 50, 60%. So automate, automate and reduce your cost going into go to market.
- [21:07]
B
So now we got, you have to figure out what you want to put in your product and you want to get those features to customers quickly with high quality and you want to have a feedback loop. And you want to be able to, since the market for these software products changes so quickly now you want to be able to respond to customer demand. So how does product management change and how does that filter through the other functions that touch customers in the company?
- [21:38]
A
Well, product management needs to spend a lot more time on really analyzing how the user is going through the product, actually invoking intelligence to help guide if there's a user of how to move through the product. But at the end of the day, the most important thing is what is the business issue that the customer is trying to address and how are they going to measure the outcomes? So if you have a outcome first mentality, hey, you know that this customer is trying to reduce their inventory carrying costs by 3x and how can AI do that? So it's an outside in versus often product management sometimes says, I know the market needs this. So it's inside out. And it's not always based upon actual customer use patterns.
- [22:30]
B
The last piece here is just the whole way you manage your financials in these companies, the whole way you price, the whole way you measure the success of your products, the whole way you manage your AI infrastructure, all that is changing and new. For the finance function, which of course this rolls through the rest of the company. But for the finance function, what financial metrics and processes are they putting in place to support an AI driven company? Because you can do all these other things, but if you're not shifting your company to align with the cost that you're actually creating and your revenue opportunity, you're not going to get there.
- [23:16]
A
Well, you don't. I now live in New York part time and I think about this phrase in gross margins, 80% gross margins, forget about it.
- [23:27]
B
So right.
- [23:28]
A
Organizations, AI, native software companies, and I'm saying the same thing for legacy SaaS, companies who want to become an AI first culture, forget about 80% gross margins. Say I'm going to model my business to know it's going to be 50 to 65% gross margins. That's just the reality of the cost for inference and tokens. And even though people are saying, well, token cost is coming way down, the complexity of these models means more tokens are being consumed per action or per task. So just plan on 50 to 65%. And one of the ways to actually recoup that, that shows on the EBITDA line or operating income line is how do I get my GTM cost down from 40% to 25% through AI automation. So that's one frame that I think CFO should really start benchmarking and talking to their colleagues or in AI native software companies is how do you basically exchange additional cogs for less cac. So it's a cogs to CAC model.
- [24:39]
B
Let's move here to looking at a really successful SaaS company that for the last three years has been on a complete AI driven mission and that's HubSpot they've gone through beginning and starting in January of 2023 they had a bunch of simultaneous and immediate priorities. Refactor their platforms for agentic architectures. Build solutions that really provide deep domain intelligence based on the data that's in their system, not just workflow automation. Train teams on AI first workflows so that their customers can maximize the use of AI. Rebuild their pricing around value and usage and outcomes rather than seats. Restructure revenue and expense budgets like you were just talking about to aligned with the new reality of how you deliver, actually deliver AI. They started early with this playbook we've been discussing but how's it worked out for them? Ray? Since late 22, beginning of 23 it seems like they've done a lot of work and they've made a ton of progress.
- [26:06]
A
Yeah, I think HubSpot is one of the better stories especially for a billion dollar plus. And you can just look at their continued eating into the market share of Salesforce in in the Salesforce automation space of CRM. But let's just. I'll do like four quick things. So they entered 2022 with a very fully planned roadmap for 2023 built around their traditional SaaS features. AI existed in small ways but they were incremental capabilities. So within weeks. So November 30, 2022 was when ChatGPT was first introduced to the world. Within weeks they basically completely scrapped their entire 2023 product roadmap and reoriented the entire company around an AI first strategy. They reallocated a significant number of engineers and product management resources towards how to use generative AI and agentic AI in their product. In fact they shipped their first AI native products like Content Assistant and Chat spot and under 90 days. So they reposition itself early as an AI empowered customer platform while maintaining their 20% revenue growth in their EBITDA. So that was one thing. They went all in and they went in quickly. They also isolated AI features so they had some platform wide AI embedded across every hub. An example, they introduced HubSpot AI as a platform umbrella and it had copilots. You know we all talked about Co pilots two years ago they had new decision intelligence they called AI Insights. And then they did some new task automation using agentic AI agents and they embedded that across their entire platform for marketing, sales, their content management and operations. And then underneath simultaneously they were refactoring and re architecting their application infrastructure with an agentic architecture. So AI number one, a lot of new products, a lot of new internal focus and a whole new architecture. And part of that was they thought of it as we're going to position HubSpot for a results of a service future versus a software as a service company.
- [28:23]
B
And they organize their company differently too. It looks like, you know, they didn't continue linearly adding employees. They figured out how to use AI to their benefit to maintain their financial performance while they were doing this.
- [28:41]
A
You're right. And some of the benefits is like to get their engineers to be really AI. First 95% of the engineers are using AI coding tools. Today. Over 35% of all support tickets are fully resolved by AI and they're trying to get that to above 50%. And the last thing I'll say because we're coming up on 30 minutes here is they did start changing their pricing fairly early on. Now they didn't do a wholesale, let's go from seat based pricing to just consumption, but they started introducing their AI products for free. Then they started charging for them on a usage basis. Then they took some of their traditional products and unbundled. So some of the things to be a smaller subscription fee but a higher consumption or usage fee. So they knew that they couldn't transform their pricing overnight and it needed to be a gradual process over time. So HubSpot in the newsletter that we published, I believe it was on Monday, February 2nd. I know we go into a lot more detail there, Peter.
- [29:45]
B
We do, no question. So let's just sort of wrap it up here. I think what we've really highlighted here is the traditional SaaS playbook just a doesn't work anymore. It's not a question for SaaS companies whether they can win in AI, it's a question of whether they can move fast enough. They need to ensure that they don't lose their structural advantages that they built up over the years. But they need to basically refactor those advantages to take care of AI. So to take into account AI. So those economics, as you said before, lower CAC account for the increase in cogs. If you're going to sum something up here Ray, for the CEO of a SaaS company making that transition to be AI driven. What would you tell that CEO and what, what encouragement would you give him?
- [30:46]
A
Well, first of all, I would say this is the largest software infrastructure transition in the history of software. Truly believe that. I also think it's probably the biggest economic and societal impacting revolution since the Industrial revolution. So I would say, number one, be really honest. Are you a system of record? Can your customer live without you? If the answer is yes, then go all in on all the things we talked about earlier. If you're a point solution, if you're going to have a real hard time maintaining your position, then think about how do I become a 10% growth and a 30% EBITDA company? Because that's not so bad. If you can generate a really healthy cash flow, your investors and you might just be happy with that, but it's going to be really hard to compete against the AI native Super Flyers. And if you're an employee out there, I think you need to learn a new AI tool or a new AI process every month, at least every quarter. You need to use that to enhance your resume. And you need to look for those companies that truly are AI first and are embracing the future. And I hate to say it, there's going to be a lot of orphan SaaS companies out there. And the best thing you can do is, number one, know if your company can be a system of record in an AI first culture or where does the next step in your career take you? And I know that is a little bit of, oh my God, it's a hot take, but it's truly what I believe.
- [32:19]
B
Well, I can't disagree there. It's, there's that phrase, may you live in exciting times when it comes to AI. That's definitely the case.
- [32:28]
A
Right? And encourage everyone to, hey, take a read of the AID RI newsletter that we launched on Monday. And that's at AI to roi.substack.com thanks for listening and thank you, Peter.
- [32:40]
B
All right, you bet.