Is SaaS Dead? How AI Is Breathing New Life into Old Platforms

Business, business, business, business.

Ah, that sounds all right.

What happens when AI meets your SaaS stack? During a panel discussion that we had at GTM Live, industry leaders tackle one of the most pressing questions in tech today. Is SaaS dead or evolving with AI? It was so fun to be part of this panel with Pete Kazanjee, co-founder of Atrium, Sean Kilpack, VP of go-to-market strategy and operations at Gong, Aditya Gautam, tech and ML lead at Meta, Louis Polan, principal advisor at Quantum Fused AI, with moderator Bala Balabaskaran, Fullcast co-founder and CTO. This was such an exciting panel for me because we talked about how AI is transforming legacy SaaS platforms and CRMs, the critical role of data hygiene and clean data in AI workflows, using large language models for synthesis, content generation, and automation, and so much more. We're really excited to share this with you today.

All right. OK, now we have our final session before lunch, and it's the one— that one of the ones that— I mean, I love every single one of these sessions. But this is a really hot topic right now. Is SaaS dead? Is it dead? How many of us are in SaaS companies? Me. Yeah, there are a lot of us in SaaS companies. So we're going to talk about if it's dead, if it's dead or not. And if it's not, why not? And if it is, how we need to pivot. So I'm super excited. Where is my co-founder Bala? Bala, come on up here. Bala is the person I was calling out that was too shy to come up here on the stage. He's amazing. Okay, Bala, I'm so excited to have you moderate this panel. And if we could have up here, we are going to have as the panelists Pete Kazanjee, co-founder of Atrium. Come on up, Pete. Woo! Louis Poland, Quant Infused AI, coming back. Sean Kilpack, go-to-market strategy at Gong, come on over. Gong's an amazing company. And Aditya Gautam, a tech ML lead at Meta. Everybody give them a round of applause. Woo! Great. All right, so Bala, you're going to be Who are you going to be? Are you going to be Jimmy Kimmel? Are you going to be Nate Bergazzi?

I just don't have the hair for it.

All right. We have one that we'll just have to pass back.

All right. This is an interesting discussion, something that keeps me up in a role that I'm in. Is SaaS dead? I think earlier people were talking about it, but is anything really dead or evolving, right? And I think that's— so the topic we're going to be exploring today is what is the impact of AI and what are the changes that we're seeing with SaaS and how is the delivery of SaaS changing with with AI. And all of you have been part of SaaS companies and you've driven that change. So we'll start with Pete, maybe a quick word about you and then we'll dive in. Okay.

Hey guys, my name is Pete Kazanjian. I'm one of the co-founders of Atrium. We make sales performance management software and are now part of the Fulcast family.

Hey, Sean Kilpack. I lead go-to-market strategy and operations at Gong.

Hi, Aditya. I work at machine learning in Meta, and I work on the ranking recommendation and how do we actually use LLMs and fine-tune and build them. Yeah, agentic AI.

And Louis Poulin, COO and co-founder of an AI startup called QuantumFused.

All right, let's start with you, Louis. Tell us a little bit about your company and what do you do?

Yeah, so by virtue of a lot of problems that myself, my co-founder, Matt Tippery, and others that we've worked with over the years, we saw a whole lot of problems with just data, data hygiene, data cleanliness, data usage. And it was always annoying in the pre-AI world. Now it's absolutely mission critical. If you don't have clean data, it doesn't matter what AI you apply against it. It's garbage in, garbage out. So one of the things that we are trying to do with our product is make this journey easier to maintain clean data hygiene, to be able to have it be optimized and maximized for usage, to identify new signals, to validate what you've already captured, and really help people think differently around opportunities that exist within their core data.

So this is an interesting question to explore in that SaaS— most SaaS software tend to be sort of monoliths, right, in terms of data with a bunch of processes on top of it. But with AI, we're seeing that get broken down and the data democratized, if you will, for agents to be able to utilize and whatnot. And that makes, as you said, data such an interesting and an important aspect of how SaaS is evolving. Give us some examples of SaaS software where you've seen that breakdown between sort of data and when you take out the business logic and you try to bring in AI to replace it, there are some challenges associated with that.

Yeah, I think most of your legacy CRM systems, I'm not gonna say the names of the companies, but there's a few that are rather large in San Francisco. I think, I think, I think your CRM platform has notoriously been a magnet and an aggregator of just bad, outdated data. People are inputting information for the sake of keeping the process going, as opposed to really understanding, challenging the data, and curating and managing good quality data that allow you then to get insights out of it. So I think as AI gets applied onto it, it's just magnifying that problem that has existed for years and years and years. And so I think those tools that are now introducing AI capabilities that make sense, that resonate either with RevOps, marketing, sales, CS, finance, you're seeing those tools start to rise to the surface. You're seeing tools that really get the AI, the next generation of AI, have been leading the call. I'll call out Gong. I think Gong has done an absolutely remarkable job. They scared the hell out of me when they started talking about conversational intelligence. I'm like, oh gosh, we don't have that. We need it. I'll talk about customer sentiment tracking through Gainsight. The AI capabilities that they're introducing truly amazing to be able to track customer sentiment, identify cross-sell, upsell opportunities, and be able to better manage and better have that connective tissue between sales and marketing and CS. So I think, I think the traditional CRM, as we talk about the topic here, is SaaS dead. I think the traditional CRM model is very much changing quickly. It's still the data that you're going to need, but now you need to think differently about how to leverage that data, leveraging AI.

Absolutely. I call it a mountain of lies. That's what CRMs are for the most part. And you know, you take and going back to Gollum, And all of the interactions with your customers happen in unstructured environments. And then what we do is we force salespeople to convert that into structured data, you know, exactly input it into the way that du jour sales methodology of the day. And then you slap on AI on top of it and see how that goes, right? So Pete, I know in, in your situation you were targeting people in RevOps, helping them with that very problem in the CRM. How did that, for Atrium, how did that evolution happen for, for you?

Yeah, I think there are, the way I kind of think about AI is impacting legacy SaaS solutions is, you know, large language model AI is really useful at content generation. It's also really good at content distillation and synthesis. Kind of like bucket those two things. And so as it relates to Atrium, we made performance analytics software that was— the primary consumer of it was sales managers, sales leaders, et cetera. Kind of the people who are the internal customers of revenue operations folks. And so there were kind of two major problems that we addressed using AI. One was the interpretation of, what we call the wall of charts problem. So, here's 20 tiles on a dashboard, figure out what's going on. And of course, the sales manager is like, close tab, right? You could use AI in order to synthesize that in order to tell them what's going on with the performance of their teams or what have you and turn it into human speak, which is words, as opposed to robot speak, which is math. So, that's the first thing. And I think that that model of synthesizing something for a different audience that can better embrace it is a great LLM use case. The other use case was genesis or generation, which is we kind of did two versions of that. One was generation of text in order to tell someone what's going on. Like, hey, this rep has these problems because of these metrics right here. The other was like LLMs are really good at code generation. And so one of the ways, if anyone's ever waited for a revenue operations person or a BI team to produce a new metric for them, usually the reason why is because there's someone who speaks SQL who's going to create it for you in Domo or Tableau or whatever. And so usually, you have to file a JIRA ticket against them. It's a whole pain in the ass. You have to wait for 2 weeks for it to come back. It comes back, it's wrong, all that sort of stuff. Well, really, SQL is code. And so if you can communicate in human speak, like, I'm looking for the win rate as defined by this of this theater right here and these reps, then the AI can generate that in code, which then can turn into a metric or like a tile that then you can look at on a recurring basis. So you're not like gated anymore. Those are kind of two examples of things that our customers really, really loved. And then once you do that on a one-off basis, you can also— this is kind of like where agents come in— you can do it on a recurring basis. So stuff that historically someone would log in and have to like type in the UI in order to like tell the software to do something. Now can just be happening in the background. So, whether that's after every call, get a synthesis of it, right? Like every deal, like, you know, use EBSTA as an example. Every deal is just constantly being, you know, curated and interpreted. Like what's going on with this deal? What's going on with this deal? You know, instead of doing territory recutting once a year, constantly rebalancing it, you know, using ongoing AI in order to say, okay, cool. Like these reps are light here. Those are, kind of the continuation of that synthesis and generation. So those are some examples of how the ways that like they've impacted Atrium and you can kind of see some of the ways that AI has impacted Fulcast. But I think it applies to any of your SaaS solutions that are out there. And if you see that that's not actually happening in your SaaS solutions, you should maybe look at ones where that actually is being embedded in there.

Sean, going to Gong, I think it's an incredible use case of of that evolution that Pete was talking about, right? Bringing in AI to sort of mine what is in the conversational platform. How has that evolution been? Sitting inside of Gong, how did you see that happen in your world?

Yeah, so I think it's been really interesting watching the fact that a lot of us have been saying that seeing this from CRM perspective and seeing all of this for a long time and now the market's catching up to it in a general adoption standpoint. So LLMs going live made everybody just like have the light bulb of like, oh wow, this is what everybody's been talking about for so long. Now what the, what's interesting is so you talk about the data foundation, everybody's like prepping on the data foundation. What's interesting now is, and I think this is why SaaS becomes so critical and so important, is how it gets utilized. An example that I think back to is I had an opportunity at one point in my career to teach an engineering class. And it was amazing how much people had trust in a calculator. They would just type in this number and just, oh, here's the output. Oh, calculator says it, so it's right. And it's like, well, yeah, you fat-fingered the thing, so you didn't even think about if this is the right answer. And LLMs are a calculator on steroids. And what we're watching is— and I watch this in Gong a lot with people taking it to different places or going and trying to create these creative new workflows, which is awesome. But if you don't have that foundational understanding behind it, And that I think SaaS helps kind of put some guards and other things around it. People are going down these paths and not having the common sense foundation understanding behind it. And they're just like, oh, well, this output looks good. And you send it to somebody who has any sort of understanding and it doesn't make sense.

Yeah, no, absolutely. And I think building on ML, which was kind of a foundation for a lot of SaaS companies, right? And from your perspective, Aditya, what, what are you seeing in terms of AI accessing data and, and people sort of giving up control, if you will, going from a deterministic kind of model in SaaS to something where you let the machine tell you what to do, you know, kind of model. And, you know, there's evolution there, there is trust there, there is sort of inertia in organizations as they sort of move along that path. What are you seeing in terms of your work?

So basically, the fundamental change that happened in all the technology before, and this AI is what you said is right at the point, it's non-deterministic in nature. So that means in the traditional SaaS tools workflow, you are seeing the very fixed output deterministic. It's a high confidence one. For the LLMs, if you give the same input like 10 times, you are going to get different answers. Right? And all the agent world, everything is basically non-deterministic in that way. So, um, the adoption is definitely happening. It is providing a value in terms of higher efficiency, lower cost, and a better user experience. But there has to be proper guardrails, proper monitoring and observation tools around this agent world. Do not go with the hype. I think just make sure like you have all those things set up, otherwise there would be like bigger issues that what like we have seen in the news and everything, right? So, so yeah, I think the data is definitely the crux of it. You need a— what, what I completely echo with like what he mentioned and just building the tool adoption is happening. Just we have to be more responsible, have the proper guardrails, observability, and like those kind of things to make sure We are having a good monitoring and the adoption is happening and we don't have to get into like we have to do it right away. Like, take your time. Like, just go with the small rollout and make sure that like you are putting it out in a safely and in a very controlled environment.

And then, yeah, yeah, no, I completely agree with what you're saying. I think also what we're seeing in AI right now and how it's reinforcing SaaS is there really is almost an adoption curve of where you're where you're going. I think we've all come from a background of do-it-yourself, go manually in, input the data, upload the data, API call the data, etc. That's the do-it-yourself world that now you can now plug into an LLM and get new insights in. We're slowly getting into the do-it-with-you. I want to get agentic AI information, I want to get other sources of information fed into me so I can have those as data sources. And I think we're— there's some examples of this working really well. There's others I still scratch my head over, but that's neither here nor there. Then I think where we want to get to is do it for you. We want to get to that automated feature. We're not there yet. And I would never advocate right now for a revenue operations organization to say, yep, we're going to completely go fully agentic AI. We're going to stack all of our agents together and I'm going to go to bed and not have to worry about my forecast. That's ridiculous where we are right now with the technology.

Absolutely.

Absolutely.

So here's the million-dollar question, right? You are the the product owner of a SaaS company and you have this data which for the most part is your moat and you've sort of built your business on that data and all the intelligence that come out of it, right? Now, what you have is AI agents basically that can be written up anywhere, can access the data and really externalize a lot of value that you provide as a SaaS company. And so you see this breakdown of your value proposition, you know, especially with things like MCP, and you set that up and, you know, you can build your workflow in cloud. You don't even have to log into your application anymore. So just from that perspective, any thoughts on how, as a SaaS company, as a SaaS product owner, how do you navigate that? And how do you think the evolution is sort of going to happen? So we'll start with you, Pete.

I don't think that SaaS UIs are going away. And the reason why is because there's a certain set of recurring tasks and things that ought to be done. And usually those show up in user interfaces as compared to if you think about a ChatGPT or a Claude or a Groq prompt window, it's like a Google Doc. It's just sitting there. What should I do here? I don't know. And so I think one of the SaaS solutions are very useful because they bundle together a set of things that ought to be done for a certain set of humans and allow you to access them quickly and do them you know, and do them on a recurring basis and share that information with other folks and have permissions and, and so on and so forth. So I don't think that's going, going away. And, and being opinionated in the things that ought to be done, um, is going to continue to be something that SaaS provider, like you ought to do territory planning, right? You ought to do deal inspection. You ought to set goals on leading indicator performance metrics. These are all things that, you know, different parts of the Fulcast family do. And so I don't think that's going to go away. I do think that if you as the SaaS provider don't put, don't take advantage of that data yourself in order to feed an LLM to allow those different jobs to be done, to be done faster, or previously they weren't done because of a competency gap by the user, or they were done infrequently because they were highly laborious. Et cetera. Like, if you don't do that, your customer's gonna turn on you and go to someone who will do that for you. So, one, you should invest there. And then two, in the meantime, you should allow them to extend off of that as well. And this is something that we learned very, very well at Atrium where having an opinionated set of metrics that— performance metrics that matter for a sales organization is all well and good, but then organizations have 10 of their kind of like fussy special metrics over here that they want to pay attention to. So allowing them to extend off of that. And then I think that also extends to other solutions who want to pick up that data, right? So I think that SaaS UIs and things like that are not going to go away, but guarding the data in such a way that you kind of make your customers make the unhappy decision between like, I'm going to fully turn on you and go somewhere else. Versus like, oh, okay, like you haven't shipped this capability just yet. It's maybe 2 quarters away or 4 quarters away. We're going to allow you to access it through whatever in the meantime. I think that's kind of the happy medium.

Sean?

Yeah, I think it comes down to 2 things. Can you provide a value proposition to the company? And in most cases, a lot of times that's going to come down to security. We're starting to see already data leakages and other things. It terrifies me seeing sometimes what I see reps dumping into LLMs. I think it should terrify every ops person. And so as a SaaS, can you provide security and comfort at a company level? That's a huge piece. And then go into a lot of what Pete said. Can you provide value to the individual user in a way that doesn't provides extra value to them. Part of that's the UI, part of it's the workflow, part of it is, like, can you get their jobs done and think of them for it, and that they're not having to create and do it on their own way. And so I think it's just an adaptation of how well SaaS can become the guardrails, the other pieces around this, to make it faster, easier, but safe for a company.

The interesting thing, as you were talking about it, is that you stay ahead of that curve, right? And use your data, provide the value as early as you can. Now, we're seeing customers basically go, well, if you don't have an AI roadmap, we don't want to talk to you, right? Because that's the sort of the table stakes at the moment. Now, if you ask them, would you pay for it? They won't, right? And so what this leads to, of course, is a lot of margin pressure and struggles for SaaS companies to sort of invest in AI and look at the monetization model around it. And it's very scarce in terms of the possibilities in that realm. So we talked about outcome-based pricing, we talked about like, you know, there's a lot of talk around tokenomics and how do you charge for that. What are you seeing, Aditya, in the models that you've come across on how to deal with that pressure that SaaS companies are facing today?

So basically what is happening, just a little bit of background on this, is that like with LLM, it's actually opened a space for new startups to kind of like compete directly with $100 billion and a trillion-dollar companies. We are seeing this like Google taking off with Perplexity, Even the recent example, like what we are seeing is with the Photoshop, for example, like Adobe came up with their new tools, Firefly, they have to adopt like pretty quickly. And then we are also seeing the Google coming up with Nano Banana, which they have launched yesterday, which is a very kind of like very convenient tool to do the what Photoshop has been doing, right? So I think there is definitely this pressure. But like, uh, the SaaS has to evolve with time. There would be an application layer which is in natural language coming over to a lot of SaaS tools. Like for example, CRM, you have 8 tabs. You need to have an interface where people can directly talk and get the results and everything right out there. So they have to adopt AI, uh, but if you have a data mode like what you mentioned, then you have a strategic advantage which others do not have if that data is not public. So that's where you can actually double down a lot of things and make it like more in a way where you can create another moat. But again, like, it's that startups coming up and competing with like big SaaS providers and like big tech companies, that is going to happen. And unless you have like a very strong moat, and even if you can have like a very closed ecosystem, you would still see the value coming up in the other one. So there is— it's expected to have a pressure, and them to have like evolve quickly with new AI tools on their— over to their traditional SaaS.

Louis, any thoughts on the pricing and monetization models with AI and SaaS?

Well, I think the interesting thing in the SaaS world, as we all know, it's the amount of tech sprawl that exists and the amount of spend that any organization has. My most recent role with Bill— sorry, I won't name the company. You can check it out on LinkedIn. I walked in and we had 120-ish tools, and we were paying for them each year. We had some homegrown tools that weren't scaling, and at the end of the day, just had really upset and annoyed customers that we were trying to serve internally and externally. So I think there's a lot of cost savings there, and I think really examining tools that are actually working and helping you grow and evolve your revenue growth and your scale as an organization, that's going to be critical. And those organizations that come up with with good AI products and offerings that add that value, that allow you confidently to say, we're going to get rid of this tool or that tool, and we're going to consolidate around this tool because I get that value out of it. Those are the ones that are going to survive. The ones that can't figure that out or have kind of a half-baked strategy, they'll die.

They'll go away.

Well, I think the strategy for most SaaS companies today is put a RAG agent on top of what you have, and that's your AI strategy, right? And exactly. So, what, what, how do you see SaaS as a, as it, as we see it today evolve, right? In, in terms of, in terms of the adoption, we talked a lot about the different ways in which AI tech can be adopted to deliver that value, deliver speed, agility, use cases that people haven't thought about with your data. So as an organization, what does that do to a RevOps person that's looking at this, right? And going, how do you adapt what you do in revenue operations to support a SaaS company that's going through this evolution? How do you think about that?

Pete?

When you say a revenue operations person supporting the sellers internally who are selling an evolving product, That is being AI-ified? That's right. Well, I think that's probably more on the enablement side, like product management and product marketing, product management, product marketing, and enablement. Because anytime you have times of flux where the kind of the jobs to be done that are the expectations in the market are changing, you need to make sure that, and then your product that is supply that matches to that demand. Is in flux as well. You just need to make sure that your field organization is highly clued in to the changes motivating why, right? Like what's motivating the changes and kind of what are the impacts associated with that? It's kind of like, you know, the sort of thing that you would anticipate doing when you are onboarding a rep to a product that is relatively static. Now you're having to do that more frequently. Maybe on a quarterly basis or what have you. And of course, there's less to be relied on of their historical enablement and onboarding at prior roles because all of a sudden you have a whole new set of nomenclatures and terminology that they have to get smart on. So my take on that would be that it's maybe less on the operations side and maybe more on the enablement and product marketing and product management side, and all the way to not just doing enablement, sessions with those folks, but also making sure that the product managers are having as much customer contact as possible and being sales engineers to make sure that it actually is working and delivering the value that they hypothesized when they spec'd it out.

This is a fantastic discussion, but I've been told we are keeping people from their lunches. Just quickly, thoughts that— anything that we haven't covered that you'd want to Any pearls of wisdom that you want to drop on folks before they get out to food?

Wisdom is a hard— is a really tough word. No, I'll say to the topic, is SaaS dead with AI? No, it's not. But it's changing so quickly that it absolutely feels like it's dying. I think the fundamentals that we saw in our current SaaS models and the processes that we all built to support revenue and revenue operations, so campaigns leading to opportunities leading to close leading to cross-sell, upsell opportunities. Those are still fundamentals. Now it's just the mechanisms of adding AI into that journey to make it even more effective. Really doing intelligent signal determination, automation of content, automation and out of— having people out of the loop as they're reaching out to customers to keep them engaged. And then that automated customer sentiment tracking on the back end. I think all of those new tools that have been introduced in the last year, year and a half. Those are all incredibly powerful, and they're gonna fundamentally change where SaaS goes in the future.

Aditya?

Oh yeah, I would just say the SaaS is not dead. Like, it has been 15 years since we have a mobile app revolution. We still use websites, right? So it just changed, evolved quickly, and I would say that like, yeah, it's a very competitive market. It gives a lot of opportunity for newcomers, so either you have to evolve really quickly, have more that AI adoption and bring more efficiency price down. Otherwise, you will be just eaten up and kind of like became dead over the period of time. So yeah.

Close this out, Sean.

Yeah, I agreed. I think it just pushes us to evolve. SaaS is no longer about access and visibility. It's about driving to outcomes. And the expectations have just been raised up.

Thank you very much. That was very, very, very interesting. Entertaining for me to hear your thoughts. Thank you all. I think you are released for food.