As AI reshapes the GTM landscape, revenue leaders share how winning teams treat their tech stack like a product—building what differentiates, buying what doesn't, and governing it all like their pipeline depends on it.
How many companies sell for millions or billions of dollars, only to disappear as products shortly thereafter?
Think of Drift, which sold for $1.1 billion and was effectively gone within 18 months. Or Segment, which sold to Twilio for $3.2 billion and is now largely embedded within a broader platform.
These aren’t fringe cases, they’re indicators of what’s happening in the market. We’re entering a phase where standalone point solutions are no longer the endgame, they’re raw material. And the winners aren’t the tools themselves; they’re the platforms that absorb them and the operators who know exactly what to build versus what to buy.
To unpack what’s driving this shift, we sat down with Kyle Norton, CRO at Owner.com, Jen Igartua, CEO at GoNimbly, and Ross Rich, CEO of Accord, for a live RevTech Stack session to bring together perspectives from operators actively building, scaling, and rethinking how GTM tech stacks should operate today.
In this recap article, we’ll explore:
Let’s dive in!
What’s happening in revtech right now isn’t random. And it’s not just a normal market correction. It’s the result of three things hitting at once:
For a long time, revtech growth was pretty straightforward. Sell more seats, expand accounts, and ride category momentum. But, guess what? That playbook is expiring (RIP).
Companies like SalesLoft and Drift were acquired at strong valuations because the timing was excellent for those companies. But the organizations operating independently today are dealing with a very different reality. Seat growth is slowing and pricing is getting squeezed. What’s more, there’s no obvious next “unlock,” so a lot of these businesses are stuck.
“It’s either consolidation or specialization that’s going to make you money, and right now, we’re definitely in the consolidation era,” says Jen. “Everybody’s building each other’s features. You used to need Outreach, you used to need Clari, you used to need Gong. Now they’ve all built into each other, and you have to pick one.”
At the same time, AI is starting to disrupt the economics underneath all of this. And not in the way people on LinkedIn make it sound. Most companies haven’t fully adopted AI yet because they don’t have the internal expertise, clear use cases, or the time to rethink how work actually gets done.
But that *almost* doesn’t matter because the shift we’re seeing isn’t being driven by what companies are doing today. It’s being driven by what investors and operators know is coming. The reality is, in the very near future, there’s going to be more automation, and more work handled by workflows rather than people.
In conjunction, AI is unbundling the tech stack. Larger platforms are consolidating and absorbing functionality, and a new wave of AI-native tools is popping up to replace those same workflows faster and cheaper. So you end up in this strange moment where consolidation and proliferation are happening at the same time.
As a result, some founders are opting out entirely because the math no longer makes sense. When teams realize they can rebuild in a weekend what took years to create, it becomes hard to justify continuing to invest in something that is quickly losing value.
And these founders aren’t alone. We mentioned Drift in the intro, but it’s not just Drift. The same pattern is playing out across the stack with companies like Pocus, Apollo, Highspot, Seismic, Showpad, Tin Can, and countless others.
The market is collapsing because companies haven’t caught up yet. They’re still experimenting and figuring out where AI fits in. They’re using AI-powered tools, like ChatGPT, in surface level ways. At the same time, the people making capital allocation decisions aren’t waiting for that to happen, instead, they’re underwriting what the market will look like once it does.
“Everybody can look six, 12, or 24 months into the future and understand where we're heading,” stresses Kyle. “We're heading for a world that is lower seat count, much more efficient, and things are done with workflows and tools versus people. And as a result, many of these companies are going to feel it. You’ll see seat compression, churn, and pricing pressure. Maybe you don’t fully churn off a tool like WorkRamp, but you’re also not going to keep paying $80,000 a year for it.”
The easiest way to think about build vs. buy right now? If it’s your intelligence, you build it. If it’s infrastructure, you don’t.
The workflows that actually differentiate you, like pre-call research, outbound personalization, lead prioritization, and anything that depends on how well you understand your customer, should not live in a vendor because those workflows are part of your GTM motion that compounds over time.
Everything else, like CRM, conversational intelligence, CPQ, etc., is a different story. These workflows are not places to get creative because they require stability, governance, and scale. The reality is you’re not going to outbuild platforms that have spent years solving for reliability, and more importantly, it’s not where you win.
Where teams get into trouble is trying to split the difference, but this rarely works because you end up taking on real engineering complexity without building something differentiated enough to justify it.
For example, rebuilding something like conversational intelligence might sound exciting until you realize what it actually involves — high volumes of call data, speaker tracking, search across accounts, and all of the infrastructure that sits behind it (yikes). For most teams, the juice isn’t worth the squeeze here.
Asana built their own CPQ over two years because their partner model was so specific that nothing on the market worked. But that’s a very different situation than most teams are in. Think about OpenAI — they’re worth roughly $840 billion, and they still chose to buy Nue, a CPQ tool.
Jen’s team at Go Nimbly built a quoting tool using vibe-coded prototypes, and at first, it worked. But the moment they tried to roll it out across users, things started going haywire. Quotes disappeared, buttons moved, and the entire experience fell apart. That’s a gap people underestimate. It’s one thing to build something that works for you. It’s another to build something that works reliably for everyone.
“Things that are the core intelligence of how you go to market that are extremely differentiated, you want to build. And things that are undifferentiated but require stability, reliability, governance, those are good product categories to buy,” says Kyle.
Salesforce is looking a lot different than it did a few years ago; it’s becoming less like a system reps work in, and more like the layer everything runs through.
And when you look at the Momentum, Qualified, Bluebird, and Informatica acquisitions, the direction is pretty clear. Salesforce is doubling down on owning the data, controlling how workflows run, and becoming the system everything feeds into, not the one reps spend their day in.
“I’ve gone from fairly bearish two years ago, I’m like, man, this is the time that Salesforce is finally gonna get disrupted,” says Kyle. “And now I’ve become a Salesforce bull because actually I think that’s the core centerpiece technology that everything can plug into.”
The reason behind this change for Kyle is simple. Even in an AI-native world, you still need a stable core. You need a place where permissions, roles, and data flows are managed cleanly, and for most companies, Salesforce already owns that layer. What’s changing isn’t whether it’s used, but how it’s being used.
Instead of trying to win as a point solution, Salesforce is leaning into infrastructure. You can see it in how the pieces come together. Momentum brings conversation intelligence, Qualified supports pipeline automation, and Informatica strengthens the data layer. Taken together, it’s less about competing feature for feature and more about positioning Salesforce as the system everything else runs through.
That shift also highlights where competitors left openings, especially in how data is handled.
Gong is probably the clearest example. By keeping data locked inside its own environment, it created friction for teams that wanted Salesforce to remain the system of record. Kyle shared that his team ultimately churned off Gong because they couldn’t write data back into Salesforce the way they needed to. That forced reps to stay in Gong instead of centralizing everything in one place, which broke how they wanted to operate.
Kyle’s team moved to Momentum, which Salesforce then acquired. And now the expectation is that Salesforce will use that to push back hard into the category.
On the other side, Gong is making the same bet in reverse. With products like Orchestrate, the goal is to make Gong the place where work happens. If reps never need to leave Gong, Salesforce becomes more of a backend system. So now it’s a real question of which surface wins. Where do reps actually live day to day?
Imagine opening an account and seeing something has changed, maybe the type flipped from prospect to customer, but there’s no clear trail showing what actually triggered it. A lot of these systems are writing back through shared credentials or integration users tied to a rep, so you lose visibility into what made the change in the first place.
This is exactly what’s happening as RevOps teams build more of their own workflows. They lose visibility and control over what’s happening behind the scenes in their systems.
“Now the issue that we’re feeling the pain of is, how do you manage all of these workflows that are sometimes opposite of each other?” asks Jen.
And it only gets more complicated as you layer in more tools. When N8N, Clay, Gong Orchestrate, and Salesforce Flow all have write access to the same records, you’re no longer working from a clean system of record. You’re dealing with multiple systems acting on the same data at the same time, without a single source of truth.
There’s also an element of risk associated with this as well — but maybe not in the way you might think. Most people think about AI security in terms of data leaking into something like ChatGPT, but that’s not really where the issue is showing up.
“The actual risk is when you use a system to create an agent, and that user has access to everything,” says Jen.
She shared a real example of this. “I saw a case where an agent was connected to Gong with access to all calls, and it was giving summaries of conversations users didn’t actually have access to. I could ask for a summary of Ross’s investor calls and get it, even though those were marked private, because it was running through an admin user.”
This shift is creating a new gap, not in tools, but in how those tools are and should be governed.
The teams that are really seeing pipeline from AI right now aren’t using templates or generic scoring models. Instead, they’re building their own intelligence and putting it directly into the rep workflow at the moment it matters.
Here’s what that looks like in practice:
That change alone drove an 87% increase in outbound call volume. It removed the two to three minutes of research between calls. When you pair that with routing reps to the leads most likely to answer, connect rates go up and so does output. Reps went from one to two meetings a day to three to four.
Of course, giving reps more time doesn’t automatically create more pipeline. If anything, it just creates more room for inconsistency. The teams seeing results are pairing the intelligence layer with enablement. They’re not just generating better inputs, they’re training reps on exactly how to use them.
“The difference between a company that is AI native and how it's building its GTM tech and one that is not is really existential. On a per rep basis, we are three to four times more efficient than our direct contemporaries. That means every rep is producing four times the ARR at the same cost,” explains Kyle.
RevOps isn’t just a workflow function anymore, it’s becoming a product function.
The teams that are moving fast aren’t just managing fields or building reports. They’re owning a roadmap, building internal tools, and thinking about the seller experience the same way a product team thinks about users. That means thinking beyond workflows and into things like:
But this is easier said than done. “Whatever RevOps team you have today, that’s not what it needs to look like,” stresses Jen. “If you run at this, you win. And you can have RevOps truly be a revenue producing function if you nail this mindset around building, owning, and operating your GTM systems like a product.”
Kyle shared what this looks like when it’s done well. His team built an internal tool called Cerebro that takes Slack conversations and Agile tickets, structures them into Salesforce requirements using Claude, pushes the build into sandbox for testing, and then ships it to production. Work that used to take days now moves four to five times faster.
There’s also a role that’s starting to show up much earlier than it used to, and it can’t be outsourced. Call it GTM tech architecture, business systems, whatever you want, but it’s not someone clicking around in workflows. It’s someone responsible for how everything fits together, including:
If the intelligence layer is where differentiation lives, you need people inside the business who understand the motion deeply and can build around it in real time. No external partner is going to have that context.
For example, the global GTM Ops leader at GoToMeeting runs a full product team inside the GTM tech org, with ownership over roadmap, architecture, and build decisions across the stack. That’s where this is heading.
Most of the teams pushing this forward are SaaS companies selling to other SaaS companies. Outside of that, a lot of industries haven’t moved yet. Manufacturing, healthcare, more traditional sectors are still early. Which means the gap between teams that are doing this and teams that aren’t is actually wider than people think — and that’s the opportunity.
“If you don’t have the right leader in place, you don’t get any of this cool stuff. You need to go push for it,” says Kyle.
The main takeaway? This isn’t theoretical anymore. The teams that are coming out ahead are making very deliberate decisions about what they own, what they outsource, and how their systems actually work together.
Here’s where to start:
The intelligence vs. infrastructure split is quickly becoming the line between teams that are building real advantage and teams that are just maintaining tools. This shift isn’t happening to revenue teams. It’s being driven by them, one build vs. buy decision at a time.
And the reality is, if your RevOps team can’t clearly tell you which parts of your stack contain your intelligence and which parts are just infrastructure, you don’t have a tooling problem — you have a strategy problem.
And the window to fix it is closing faster than most teams realize.
.png)
