The Adaptive GTM Model

Julia Nimchinski:

Next up, welcome, Seth Marrs, Chief Strategy Officer at Sandler. We have an all-star CRO panel here. Welcome, Seth. How are you doing, and how is London?

Seth Marrs:

Good, good. Always good being Lebanon. How are you?

Julia Nimchinski:

Awesome, super excited for this. Welcome, Stevie, Niroshan, and we are bringing everyone on here. Take it away, Seth.

Seth Marrs:

Alright, so, this one’s a particularly interesting panel, so I’m really looking forward to talking to this group about it, because I’m a very, very, public skeptic on the next best action when it comes to B2B sales. I just look at it and see… the complexity of these sales, and that it currently requires seller’s judgment, intuition, to really understand what’s going on, and that you can’t just prescribe a next best action, and have a seller follow it, and be confident that that’s going to work consistently in the long run. Now. That doesn’t mean that… that I don’t believe in the power of AI, or of system-driven insights, or that context can’t help sellers.

I just think it should be a set of options rather than a, here’s what you should do, and that’s really what we’re going to spend this time talking about. And the context of it is really talked about in the adaptive go-to-market model. That’s kind of how I’ve… defined it, so we’ve got a great group on. I’m gonna jump in, and the first thing I’d like to do is just have each of you get started by saying a little bit About yourselves, and then just talk about your perspective on the current maturity level of adaptive go-to-market capabilities.

So I’ll, I’ll, I’ll start first with, with Philip.

Philip Lacor:

Hi everyone, my name is Philippe Lacour. I’m the COO of Personio. We are a late-stage HR and payroll company, about 1,500 people, mostly focused on the EMEA market. On the adaptive go-to-market model. I would say that we got a lot more data and a lot more signal. And in some areas, it’s working really well, and in some areas, it’s working less well. And, one simple distinction is between existing business, so existing customers and new business. We typically see that on existing customers, we got a ton of data around how to use our product, how to use the different modules in the platform.

Whether they, they are looking on certain parts of our platform to indicate, hey, there’s potential to sell more. And then on the other hand, although we use Clay, LinkedIn, all the others. On the new business side, yeah, you can get an indication of who’s in a buying buying pattern, but it’s not… it’s not a black and white. So, there, I don’t think we got the precision yet that is needed to, to, to be perfect. We… we’re good, it helps, they’re the signals, but I think there’s still more to be done on the… on the new business side.

On existing business, we got… we’re a lot further along.

Seth Marrs:

Yeah, so it’s just kind of a… A lack of reliable data, would you say, on the net new side? Or a lack of understanding how to put it all together?

Philip Lacor:

I think the availability of data of a prospect is very different than from an existing customer, right? There’s simply way more signal, and therefore, the AI can help you really pinpoint on where to target and when to target. But for a prospect, yeah, you can see when your persona has changed job, or that they are a fan of Andy Roddick, or whoever, but that’s not going to necessarily mean that they’re ready to buy their next solution. So, I think that’s where you see some differences.

Seth Marrs:

That makes sense. It’s a good transition, James. I’ll let you introduce yourself and talk a little bit about this, too, because I’d be interested in your perspective, especially on the prospecting side.

James Roth:

Sure. Hey, Seth, good to see you, and it’s good to see several friendly faces on here. I’m James Roth, I’m the CRO at ZoomInfo. ZoomInfo is the world’s biggest go-to-market data provider, so we’ve got about 40,000 customers. Over a billion two in revenue, and, you know, ultimately. trying to either be the interface where folks do their prospecting, so we’ve got a very large business, which is our purpose-built tools for prospecting, and then I think most recently, you know, over the last 6 to 9 months, certainly, like a lot of other folks, moving towards more of the headless, piping our go-to-market data in. to, you know, whether it’s an LLM, an overarching context graph, and, you know, really trying to help with marrying the third and first-party data together.

Philip, which is a great segue, you know, I think as we see the maturity curve, you know, of our 40,000 customers, you know, I would say it’s heavily weighted towards upmarket, you know, call it 100 employees. to, you know, unlimited employees. We’ve got about 80% of the Fortune 500, and I think what we see is a pretty significant chasm. between, you know, companies that have started the journey, companies that have not started the journey, and companies that are incredibly advanced in the journey. And I see Stevie on, and, you know, love, love our partnership with Vanta, and, you know, the work that they’re doing, you know, especially in building that overarching context graph.

I think what we see Is… with the amount of… noise in the market, and obviously, you know, trillion-dollar potential IPOs with some of the really big names in the, in the sort of native AI space. You know, that was definitely the rush. It’s like, okay, board’s telling me I have to do more with AI, so I’m gonna go buy this. And then, I think where folks kind of take a step back and understand that, you know, the LLMs are amazing. There is no denying that in terms of what you know, the output of the MAR, but I think folks then come back to, okay, how do I tune this underlying context graph?

How do I make sure all of my first-party data is in? To Philip’s point, you know, tons of existing data on existing customers, you have product utilization, you have what they interact with from a marketing standpoint. You have things like who signed the DocuSign, like, all of that rich first-party data should be plugged in, because it can tell you an amazing story about what’s happening within that particular vertical, within that particular customer. And then marrying, you know, basically as much third-party data as you can into that underlying graph to get this mosaic, if you will, across all things first and third party.

And then the… I think the companies that are the farthest ahead are the ones that are constantly tuning that, piping in as much information as they can, and then tuning the LLMs to find those for their respective use cases, whether it’s, you know, expanding TAM, whether it’s, you know, scoring in terms of propensity, whether it is, you know, like an upsell or a downsell or a churn mitigation, and so I think getting that right When we see the companies that are, you know, far, far, far ahead, You know, they’ve really… leaned into that idea of, I want to have the most powerful contextualized graph across all first- and third-party data, and then put the LLMs on top of it, and they have go-to-market engineers that are constantly tuning that.

So, you know, that’s at least what we see in the space, but, you know, it is… the distance, if you will, between folks that are just starting versus folks that have been at this for a year, two years, it is pretty wild to see Just how far along some are, and how far behind others are.

Seth Marrs:

It’s crazy. But, I mean, the point’s really interesting. It’s constant tuning. There’s not a finished product in this, ever, it seems. Like, you’re just constantly working through it. Cool. Stevie, why don’t you go next?

Stevie Case:

Okay, hi, great to see you all. So many friendly, familiar faces here. My name is Stevie Case, I’m the Chief Revenue Officer at Vanta. We are a Series D, trust management platform. And we serve a wide variety of different types of customers, everything from early-stage startups, where we sell to founders and help them with first-time compliance, all the way to the Fortune 500, where we help CISOs to manage their security and compliance programs through a single pane of glass. This is an interesting moment for us. We’re a large go-to-market team. We are 100% sales-led, so we are in a position that certainly put us at the front of the pack in terms of teams that were focused on automation with AI first, and then really starting to think about adaptive go-to-market and what that could look like.

So, we started that journey early. I’ve got a team that is about 800 folks globally. That’s everything from salespeople to SDRs, account management, customer success. partnerships, channel, RevOps, systems, analytics, and that crew, because we are so human-led, we started with exactly what you would expect, and that was trying to automate away the busy work. And the beauty of that is, I think there are a lot of tools you can buy that do the basics very well. I think you can also build the basics very easily today. So a lot of those early automation wins and the low-hanging fruit there, I think that those boxes are well checked.

What that then led us to was the next stage of the journey, where we started to think about next best action, to your point, Seth, and how we start to do something a little more differentiated, and create a unique customer experience, and really drive value, and make it easier for our sellers and our customer success folks, our support team. To understand how to better serve customers, how to understand what next steps really make sense. And… we learned a lot of lessons on that journey.

You know, I think, like a lot of folks, we all started building, we gave people tools, we gave everybody Claude code, we gave everybody dust, everybody was building agents, and some of those agents were incredible and really impactful, and they existed in a silo of one person, and then whoever else they copy-pasted the prompt to, and, you know, it became this mass of hundreds of agents, some of which were really elegant and well-designed designed and fed rich data in context, and others of which were operating without context and without good structure or infrastructure behind them.

So, as you would expect, we ended up with a tremendous amount of spend and an unoptimized stack. And one of the biggest revelations of that moment, because we were building both bottoms-up and tops-down from an applied AI team, was that the context graph and everything that came with it was the piece that was truly missing. We had a lot of data. It was not sufficiently rationalized or available in a way that people could build agents and get the same answer, or get the right answer in context. So that led us to the next phase of the journey, where we really undertook a huge project led by our systems and analytics team to get all of that data into Snowflake, to build a semantic layer, to make it agent-accessible.

And that is the era we now live in. And now, we’ve gone about building apps at the core, we’ve got a system for promoting and making things the blessed agent that has been actually built and checked by our centralized team. We still do bottoms-up, we still do tops down. But now we’re in that era, we’ve actually built some apps, including a new customer success platform that is replacing the SaaS that we built, that we’re churning off of. So, we’re a combo build versus buy culture, but it’s been a real journey, and I think there’s a lot of upside still to come.

Seth Marrs:

Fantastic. I’m looking forward to going through that and some of these questions and hearing your insight. Thanks, Stevie. Graeme, you wanna go next?

Graeme Geddes:

Yes, I’d love to. So, great to join everyone here. Graeme Geddes, Chief Sales and Growth Officer with Zoom. I think most might know Zoom, from, you know, from, you know, kind of, the core video use cases, but really Zoom as a communications platform extends well beyond the video that we’re most known for, and really, we’ve extended into, you know, areas across the platform with the sales persona, right? How do we help sellers be more efficient? So, I am both a customer of technologies here, as well as a supplier, and so, you know, very excited to share some of what we’re doing there in terms of dogfooding, or even, you know, as we like to say, drinking our own champagne.

And I would say that where we’re at in our journey is very similar to what you’ve heard from others, but really it’s come down to this understanding that the context graph is so vital and important for unlocking where we can go with AI. And, one of the benefits that we have, right, is the context graph is so rich with conversational data, when we look at for our customers. just trying to understand the systems of record, right, what’s logged in Salesforce, or, you know, what might be, you know, in other, you know, kind of static databases is such a fraction of the true understanding of where your customers are at, both on the buy side as well as supporting existing customers.

You know, to the points that Philippe was mentioning earlier. And so, you know, we’re really excited about where we can… where we are today and where we can take this technology in terms of taking that conversational understanding and empowering that into the Agentic Harness loop. So that we can make sure that that is a continually evolving and kind of living body based on everything that’s happening and all of the conversations that are happening in real time across an enterprise.

Seth Marrs:

Fantastic. I’m really hoping you weave in some of the stuff that you’re doing around Zoom Revenue Accelerator. Everyone was kind of taken aback by the common room acquisition, so it’s really interesting to get… it’ll be really interesting to get your perspective there. Cool. Niroshan.

Niroshan Rajadurai:

Thank you, Seth. My name is Niroshan Rajadurai, and I’m the Chief Revenue Officer at Expo. We are an autonomous offensive security platform. In other words, we’re basically a platform in which you can use AI to hack your applications. Just sitting here and listening to all of you talking about the wonderful things that you’re doing within your organizations, I’m really excited for this conversation. Just, you know, amazing progress that you’ve all made here. The challenge that we have at Expo is that we’re an organization that’s 2 years old, and we’ve only built a go-to-market motion That’s about a year old now.

So, one of the, you know, the challenges is that we’re very new, but one of the that obviously we have a greenfields opportunity here for how we want to design this. So, from day zero, we made a conscious decision that we wanted to be AI-first in revenue. And we actually built our RevOps organization as a product organization. And so that really meant that when we looked at, like, the technology stack that we wanted to bring in, we really looked at it from a perspective of how do we scale, how do we have, like, a clear foundation?

Where we know that there’s data truth, and using that, we can then start to build all the Agentic layers, and all the reasoning layers, and all the different pieces that we can then use to go and serve the organization. So, you know, we literally started with zero data. And we’ve been building up from there, and as we kind of started spotting… started identifying blind spots in what we’re doing, we’d add more instrumentation into… into the process. And so today, we have a set of agents that basically are like oracles, right? So they’re essentially like an individual.

You can go and so multiple, engagements with the agents, you know, span maybe 10 to 15 conversation items in a particular thread. So it’s more than just, like, here’s a piece of information, it’s really, like, you can dig into the details, dig into the data. And then on the other side, from the prospecting, so just much like the colleagues here, or peers on this call, where, you know, we’re talking about, like, just integrating into different signals, we plugged into tools like Octolens, into ZoomInfo, into Sumble, to really get a pulse of what’s happening. And then, you know, I would say, like, we’re generally, as a go-to-market organization, we’re in the business of selling trust and trusted advisors, right?

Because we’re a security technology. So it’s really important that when we go to someone and present something as innovative as what we’re doing right now, that we really demonstrate that, A, we understand their business, and we understand what recent challenges that they’ve had. But we’re trying to do that at the scale of a company that’s just, you know, completed its Series C. We’re an AI-native company, so, you know, the growth expectations for AI-native companies are very aggressive. So we really have to empower our field to be able to leverage this information, be audible, ready when they have those conversations with those customers, and really hit on what the challenges are and how we can help them.

So that’s… that’s what we’re building out, and yeah, looking forward to… to continuing this discussion here.

Seth Marrs:

Yeah, it should be a cool perspective, because you’re one of those companies that’s having this really quick upswell in growth, and having to move into sales quickly. It should be a really interesting perspective. Alright, so I’m gonna jump into the first question. So… what is the new buying signal, that AI’s generated, that if you went back a year ago, that… it wasn’t even available that’s helping you win deals. James, you want to start off?

James Roth:

Sure. You know, I look at this less of a, here’s this new signal that didn’t exist before, and it’s amazing, and we’re winning all these deals with it. I think the biggest unlock we’re seeing, both internally and externally, is what the AI is really providing is the ability to amalgamate many signals, and then give the rep a chance to then synthesize that, and then use it with a thoughtful point of view. And so I think if you think of the traditional signals, whether it’s, you know, Stevie joins Vanta as CRO, or, you know, company just reported earnings, or somebody’s on my website. you know, those signals traditionally were looked at as an individual, and so, you know, an individual signal, that is.

So when Stevie joins Vanta, people see that, you know, whether it’s in LinkedIn or on ZoomInfo, whatever it may be, and then Stevie gets a thousand prospecting, hey, congratulations on the roll, can I sell you some stuff? And I think that it’s not that that’s not a powerful signal, it’s just that, that entire mosaic, if you will, of, okay, Vanta has been a customer for this long, this is their utilization, this is their trends, Stevie came from company X, and before that, company Y, this is what she’s posting about.

This sort of… collection of the different signals that you would have to hope, and I think we’ve all been in this place years ago, where you had to hope that your rep was looking at 10Ks, and they were living on LinkedIn all day, every day, and then they would go use a ZoomInfo, and you would hope that they would do all of those things and be, like, your top 5% or top 10% of reps.

Or, and to Graeme’s point, you know, the amount of just available, amazing information that comes, and we’re a big Zoom customer, that comes from those conversations, those interactions. you know, hoping that somebody would be listening to those and looking for those key moments, the ability now to take every single one of those, put them together, and say, okay, like a propensity score, that has always existed. A lot of times in a black box in marketing land that sales reps rarely ever trusted. the signals, the third-party signals, those existed too, the point of view, the ability to do that research, all of those things existed.

The ability for them to not only be across all of those signals at the right time, and then, I think we’ve all been frustrated in the past, where then the rep has those, but then they might pick the wrong spot, or the wrong point of view, or they go try to sell a company that just announced 30% layoffs. They don’t have the whole picture, and so I think the ability to synthesize all of the signals that are happening within a potential customer, a prospect, etc.

And then providing it a thoughtful point of view, and I know one of the things we’re going to talk about is signal to action versus, you know, signal to option, if you will, but… I think the ability to do that, and then decide, do I want to automate this, do I want to have an agent run this from an inbound perspective, from an outbound perspective? Where do I want the human to come in? None of that really existed without that… gonna be a broken record, but the overarching context graph and the ability to synthesize all of that information into a thoughtful point of view.

Seth Marrs:

Isn’t that the new signal, though, right? Like, before, you just had a bunch of stuff giving you, like, onesie-twosie things to go do, and maybe they worked, maybe they didn’t. For the most part, they didn’t, because every single person that was trying to sell that company got the exact same signal, so to your point, they’d all send the same email at the same time. You guys have all talked about context graphs. Isn’t that new signal that’s happening today, is if you have a really strong context graph, it aggregates all of that together and provides a signal that is well beyond anything you would have seen before, and also contextually relevant to your company in that particular deal.

James Roth:

100%. And then I think the ability to then marry that to, okay, this is the vertical, this is the sub-vertical, this is their size, this is how many customers we have, these are the moments that won the last 100 deals in mid-market fintech. you know, taking not only the new signal, which is effectively all the signal, but then marrying that, you know, I think one of the hardest things, especially hiring large teams of salespeople. you know, I remember 5 years ago, you had to go pay a huge premium to get a vertical-specific rep, somebody that had been selling to financial services their entire career.

You had to pay them a 30% premium, and really what you were paying was for that business acumen, or that business context. Where now, that business context can be basically batteries included to say, you know, back to the conversational intelligence side, okay, here’s the last thousand hours of calls with mid-market fintech between 100 and 500 million in ARR, This is who’s in the buying committee, these are the titles that matter, these were all the people that joined the calendar invites, this is basically your buying committee. Versus, I think, a couple years ago, you would say, oh, we sell to X person, we sell to X person ubiquitous across all verticals.

Not the case. And so I think pulling not only that new grouping of signals to say this is the right time to reach out, and then being able to pull instantaneously business acumen, business context, vertical context, here’s who you should reach out to, you know, at this particular vertical, there is no SDR leader, because they don’t have SDRs, they have leasing agents, whatever it may be. I think the amount of signal waste that has taken place over the last decade, where you give that information to a 23-year-old BDR that has never sold to commercial real estate before, never sold to regulated financial services before, you know, now you can give them a fighting chance, if you will, because the context is clear, the vertical is clear.

So yeah, I mean, 100% aligned with you, that is the new signal, and then I think it’s taking the new signal, and then being able to action it with that relevant context, and then to the point of tweaking, do you want that to be something that runs autonomously? Is that an always-on go-to-market play that a human never touches? is this, you know, if you can get to a signal stacking that is so strong, you say, this is what I want to go to our best folks, because I don’t want them, because, you know, I’ve got these 8 to 9 signals.

So I think, you know, expanding upon the scoring model with this additional synthesis of all of the signals, I do think that’s where a lot of folks are going.

Seth Marrs:

Very interesting. You know, Sean, you’re, you’re, like, right in the middle of this, so you may have, like, for this business, just completely skipped the whole onesie-twosie signal, and now you’re in the middle, like, are you seeing that same thing play out as you’re building this out, that you’re using aggregated, multi-dimensional scoring? for your next… for your actions. Are you seeing what we’re talking about here playing out in how you’re building things?

Niroshan Rajadurai:

Absolutely. I mean, I think, the big breakthrough over the last year in general, so, is… the ability to just crunch through a huge volume of signals, right? Typically, what you acted on previously was much smaller, just because it was just too hard to get to it. And then the second thing is, because you’re able to crunch through that much of data to decide, like, what the signal you want to act as, you can be more aggressive in your instrumentation. So if you think about, like, lead flows, for example, like, what brought someone to a particular engagement.

You know, today, we can actually track that all across an entire business to see, actually, like. the email went here, this person sent the email to this person, that person went to a webinar. We can see on LinkedIn, that person is connected to this person here, and that’s how we got a reach out. So we can see those kind of relationships, and so that gives you a much better understanding of, like, what are the triggers that actually cause a business to engage with you. And that means we can actually go back, because, you know, a lot of what we’re doing today is prospecting, because we’re building pipe, aggressively.

Like, we 3X’d our revenue organization in the space of, like, 3 months. So, pipeline building is, like, central to what we do, so the question is, like. How do we quickly understand what works. how can we understand that with a, you know, a very granular level of detail, and then how can we work with, like, the marketing team and the outbound teams to, like, recreate that playbook and go back and actually have, like, drive more meaningful conversations, across new prospects? So that, for us, is, like, how We’re leveraging the signals to just build pipeline up front.

Seth Marrs:

And it sounds like, going back to, I think, James, you talked about it earlier, that never-ending kind of tweaking that’s happening as you’re going through on the pipeline build, 3 months is a long time in your company in terms of what you’d be looking for then versus… versus what you’re looking for now, it seems.

Niroshan Rajadurai:

Exactly.

Seth Marrs:

Alright, so… The next question is, like. So you’ve got this mature, adaptive go-to-market model set up, everyone’s got context graphs, they’re serving up all of this information. what’s the role of the seller in this world? I mean, in this kind of environment, like, yeah, how does the seller play in? Philip, you wanna start us off?

Philip Lacor:

Yeah, yeah, the role of the seller is, first of all, still very important, but it is different than it used to be. The time spent usually on information gathering has totally changed, as all of you said. We’ve all been building these context graphs. You can now build vertical specialty very, very quickly. One of the things that we do is bring even these contact graphs To the sellers. So every morning, you’ll… they will get a morning briefing with, 3 meetings for the day. It has a summary of the last pain points, the last conversations, the personas, the key questions to ask for everybody on every one of the meetings, so it’s really in their fingertips, they’re getting this in Slack through the cloud agent.

So this information comes to them in a very, very easy way. Having said that, it’s still not enough, and now you need to be able to apply judgment in a way the fundamentals of sales now become even more important. I’ll give you an example. A few weeks ago, we were working on a very large deal. And we use Matic as a qualification methodology, and the economic buyer in our sales cycles is not the HR person, but the CFO. So I tend to always ask, okay, how did you engage the CFO? How did that go? And, there was a very experienced rep, and she said, look.

In this case, we should not engage the CFO, and here’s the reason. And she explained this whole thing, and knowing her track record, I went with it. And of course, a few weeks later, they closed a deal. And that is judgment that would not have come from the context graph. That is her experience, her, like, interpretation of the environment, whether she needed to slow the deal down or accelerate it, whether I need to be pulled in, or our CEO, or not. And those things is what still set the best sellers apart. From, like, everybody else who has the same context data.

Seth Marrs:

The level’s going up, but that judgment piece is still the difference maker.

Philip Lacor:

Yeah, absolutely. You gotta, you gotta be able to read the room and feel what’s going on.

Seth Marrs:

Absolutely. Graeme, how about you?

Graeme Geddes:

Yeah, I would say I agree with that, and what I would say is the role of the seller is changing, but I would argue it’s actually becoming more important, not less. And so, you know, first, and kind of foundational, is making sure that you do have, you know, AI and Agentic systems that are offloading a lot of the, you know, the research, qualification, routing, follow-up, meeting preparation elements of the sales job. And… Now, sellers are really, you know, being involved when it… requires judgment, right? How do you build trust? How do you understand influence within the organization, and kind of navigate through that change management piece?

So, you know, I think that the days of the past might have been, right, it was the seller who knows the most information, and I think the comment that was made before, right, whether you understand that specific vertical, right? Now the system is going to understand more information than the seller. So the seller of the future is the one that who can take that intelligence, right, and turn that into an actual customer decision.

Seth Marrs:

And do you think over time, I mean, if you have great sellers that are making those decisions, and those decisions are being captured, and then fed back into the system, does that make the seller a really, really good seller more valuable, because you could democratize their intelligence on these deals across your entire team?

Graeme Geddes:

Yeah, I would completely agree with that, and we’re actually already seeing it. We’re seeing, right, the best sellers are, A, the ones that can leverage these tools, and that’s differentiating them versus their peers, and then secondarily, it is that feedback loop Right? That’s taking, you know, okay, what are the outsized outcomes that we’re getting from a small cohort of salespeople, and then leveraging those best practices and having a true understanding of every point along that customer journey. What are those deciding factors? And then kind of redeploying those across the rest of the organization.

Seth Marrs:

Fantastic. Now, Stevie, you mentioned at the very beginning that you started sales first, and you’ve kind of pulled this in where you’ve now pulled the context draft in, and have really started to integrate this into a Salesforce organization. I’d really be interested in your perspective on this.

Stevie Case:

Yeah, I mean, I would say, if anything, this has made the art of sales more important, it has made, like, being a human being with judgment more important. I think that, you know, everything we’re automating here, all the signals we’re feeding people, the tools we are giving people, it does allow you to meet kind of a minimum productivity baseline pretty easily. I was actually just at my desk talking to one of our SDRs, and he said, you know, the reality is. you’ve handed me great tools and this amazing stack, and if I just sort of go through and check the boxes, like, I can hit my targets.

But man, I’m really in a position, if I apply judgment and creativity. to blow it out of the water and deliver an incredible result. So I think what’s now really the opportunity is to be human. It is the judgment, it’s the relationships. People still buy from people. If you look at what’s happened with all the slop that fills all of our inboxes. it’s made email completely useless. Like, I don’t… I actually used to occasionally respond to cold emails that were really, like, artfully written, and… absolutely never respond anymore. So now, and we see this on our team, the thing that’s working, it’s cold calling.

It’s human beings picking up the phone and having a real conversation. So that is one reality. I do think it makes it harder, I think it makes the art more important. The other reality here is the system is not static. The ball moves, the game moves, the data and the signals move, so… you cannot build a system, deploy it, and just let it run. You have to be constantly iterating. The context changes, and you can build systems that iterate and learn, but you still need that human judgment on top to help with iteration and to make it unique, because If you’re just building a system that is pure automation and no human judgment, you’re gonna end up with something that’s very commoditized and common and that anybody can replicate.

So, that human judgment will ultimately be the thing that sets certain teams apart.

Seth Marrs:

It’s getting from the average. The bar’s going up, but to be better than average, you need to have that seller that’s pushing beyond what it… what everybody knows or what everybody sees.

Stevie Case:

That’s exactly right.

Seth Marrs:

Fantastic. Okay, so I wanna… I wanna continue with you on this, on the next slide. If we go back to, like, you talked about the context graph, and we… we talked about all of that type of stuff. The main thing talking in this, in this, event is around the harness, and how… how do you currently see… or define a harness to your team, or do you even talk about it? Because now all the tech companies are talking about the harnesses they’re building, and those types of things. How do you talk about it, and how do you think about it from what you’re doing in terms of building a sales technology structure for Vanta?

Stevie Case:

Yeah, it’s interesting, I mean, we talk about it in terms of, like, the harness we’re building inside of our core products, so we talk about it with our customers. We don’t talk about it so much in terms of what we’re building inside of our go-to-market organization, at least not yet. I do think it is inherent in the conversation, and the way we’re architected now, so we do still have a very bottoms-up culture in that we’ve given everybody platforms that anybody can build agents, we’ve given them now a system to promote those agents and have them become blessed and refined.

And we’ve given them real systems access. We actually opened up write access into Salesforce, so we’re giving people more access than ever before. We do also have an applied AI team, and they are our kind of tops-down, central core build team. Really, a lot of what they’re doing is setting the architecture, making design decisions. and really guiding, like, final build, so they’re like a product organization inside our org. So, we are talking about Harness and evolution of Harness, but without using those words. You know, I think it’s been more within go-to-market about what has been the platform that we want to really get everybody focused on?

Originally for Bottoms Up, that has been dust for us. I think we found our more hardcore builders are actually strictly in Claude Code, and then our Applied AI team is primarily Claude Code with some other kind of unique bits thrown into the mix. And I think that’s going to continue to evolve, and, you know, I think what Harness means in that context really hasn’t settled yet. I think it’s a clear conversation if you’re in engineering and you’re in the part of the world building the core products, but for folks like us who are building, I think we’re still settling, you know, what it means and what it looks like in go-to-market.

Seth Marrs:

Yeah, whatever I need to do, call it what you want, I’m just trying to find a way to get results.

Stevie Case:

Yep, that’s exactly right.

Seth Marrs:

Niroshan and Philip, you two both are in a similar position to Stevie in that you’re selling a product to… you’re not one of the tech companies. How do you guys see it? Philip, like, where are you at? Niroshan? I’d be interested in you, is it kind of where you are compared to where Stevie is.

Philip Lacor:

somewhat similar. We gave everybody at the beginning of the year access to Claude, so everybody started building agents, so we got tons of agents, it was all great. And then we were like, okay, we gotta, like, start, like, standardizing. I mean, all of us spent, like, many, many years building best-in-class processes, right? We standardize our forecasting process, we standardize our sales process, we standardize how to put a proposal on the table, and all of a sudden, we let all of that go and say, hey, go build and go figure it out, and that’s what we got.

And what it did was, it did two things. One is, it drove a lot of adoption in AI. well, it did 3 things. It drove up, creativity and engagement of everybody, and it drove up token cost, and And then we were like, okay, hold on, We have been working… there’s a reason that we’ve been working so hard on all these processes and best ways of working over the last year, so let’s start bringing that back, so… We have a central go-to-market engineering team, we have a central data and system and AI team, and we now have, for every core function, like BDR or account manager or CSM, we have a hub, and that is the agent hub.

For example, AEs have 8 core agents, and those are the… let’s say the power agents. And, it’s interesting, we play back that, some people were, try… had their book of business and asked Claude, okay, show me what the best accounts are. And we actually demonstrated that, not only was the token cost higher, but also the results were infer… inferior to the pre-built models that we have been fine-tuning over months and quarters. And I think that’s important awareness for the team, because they go like, okay, let me just pose a question in Claude and get the answer.

But you don’t get necessarily the right question, and although we don’t use the word harness internally. It’s important that you say, hey, these are the best agents, where we really fine-tuned the contacts, the behaviors, and those are the ones you gotta use. And if there’s new ones, then we’ll look at them and scale them up if necessary, but that’s how we’re gonna do it. And it’s more like 80-20 or 90-10, and over time we’ll get better and better, but you gotta get back to some standards.

Seth Marrs:

Yeah, you can contribute, but you need to contribute into the main model, not your own.

Philip Lacor:

Yup.

Seth Marrs:

Got it, got it.

Niroshan Rajadurai:

Yeah, I mean, just to follow on there from Philippe, we have an unfair advantage, because Expo is in itself a Harness framework around a bunch of agents, and are actually, you know, designed to run for multiple weeks at a time on solving a problem. So, one of the things that we tried to do when we set up our environment is really, like, how the harness works, all those semantics, we basically hid from the field. Our job was to make sure that when you asked a question, there was data integrity. as the agents went and did the work to fetch that information or do actions, they don’t do anything, you know, nefarious or bad to the system or corrupt things.

And we basically provided those foundational controls. And then what we tried to do is actually humanize how the GTM team and actually the whole business then engaged with what was going on. So we provided them a single interface, which was this kind of multi-purpose bot that sat on Slack. And whether you wanted to do prospecting, whether you wanted to understand, like, opportunity, MedPick alignment, next steps, whatever it was, there was actually a single point of control, a single point of engagement that the field worked with. And so what this meant was, we really got everyone focused on, hey, you’re here to sell. you’re here to support a customer, you’re here to, you know, understand the technical needs.

I’ll just give you an interface to do that. If you want to go into the backend and contribute, you’re more than welcome to do that, but we all know that what you’re really here to do is do your core job. and support you and do that. And we just really, really simplified that interface, and then made very sure that every time we presented information to whoever was asking for it, it was accurate. So there was always trust. And so the key part of that is having the harness and the controls in place to be able to do that, but then abstracting all that complexity from the rest of the team.

And that’s worked out really well, and so that’s become a model now that’s scaled from legal, for example, has their own specific bot for legal. and Human Resources and Talent has their own specific bot. But for the purpose of GTM, there’s a single interface that is a humanized interface in the way that people are used to engaging with another person. They can engage with this agent and actually get the questions answered and have actions taken from what they want.

Seth Marrs:

Interesting. So it sounds like, I mean, in all three of your cases, you’re building harnesses for your individual business. Like, Graeme and James, like, you’re selling sales technology into these types of companies, and the harness is at the center. There are many companies that talk to the harness as a foundational part of what differentiates them. Can you talk a little bit about just how do you position that in a world where you have each one of your customers building their own internal harness for their business?

Graeme Geddes:

Yeah, maybe I’ll start there, and I’ll say, you know, back in June, we launched ZoomMate, which is our, you know, our harness, right? So our agent technology, and you can think of it as an agent of agents. So, you know, we believe very much, right, that you should ask the question, and then ultimately the harness should be the orchestrator of multiple different sub-agents. Within ZoomMate, we have both workflows and skills, right? So my team has kind of taken off, you know, to the races in terms In terms of developing different skills, where we publish, we support, you know, the best ideas across the team, the best workflows, and then ultimately, those have now come and been productized around the sales persona that we can then launch with default skills and workflows for our mutual customers.

But I think the important part that you’re mentioning here is ZoomMate is available both in terms of a first-party, or excuse me, kind of a first-person harness, but also headless. So, the rich contextual understanding of all of the data, both systems of record, but also the systems of engagement, the conversational graph. Right? We can plug that into a customer’s harness, whether it be, you know, Codex, or Cowork, or an internally homegrown solution, so ZoomMate can participate and add that additional kind of context layer. So, the answer is, is we’re meeting our customers where they’re at, but in terms of internally, we’re using ZoomMate, you know, as that harness across the go-to-market organization.

James Roth:

Yeah, very similarly, last summer, so July of 25, we built our internal, we called it Mesh. It was all built on Claude, it’s got all of our first-party data, all of our third-party, and then waterfall across all the other third parties. And it was one of the most heavily utilized products we’ve ever built, and I think one of the things that we have, which is, you know, it’s a great place to be, is we’ve got a very large go-to-market team, and anything that we roll out, we get to learn a lot.

You know, we sell to salespeople, we sell to marketing folks, we sell to RevOps people. we are customer zero for whatever we do, and I think, you know, that was the first foray into, okay, this is what large enterprises or forward-thinking or sophisticated customers are going to build on their own. They’re kind of done with buying purpose-built SaaS applications, so we built that internally. We saw the exact same thing. We had thousands and thousands of agents built, probably 10% of them were great. 90% of them were just worse versions of a lot of the same things, and so we could start building the agent library, we could start really saying, you know, I didn’t want account managers to be great agent builders, I wanted them to be great at using great agents.

And so we built out the same applied AI team internally. We have a purpose-built go-to-market AI team, and then we have more of a overarching, for the entire company, AI team that built Mesh. And so, we got to learn a lot from that, and then as we’ve been building products, you know, it’s basically, you’re gonna have a cohort, to Graeme’s point, you’re gonna have a cohort over here that will just build everything. You know, I was meeting with the CRO at Vercel, he’s like. Literally never gonna buy a piece of software again, but we are huge consumers of data, so great, that is a build, and we want to be as easy to pipe in our data into that, you know, what they’re building.

And then you have customers that might have a RevOps team of 2 or 3, they might have… be wearing a thousand different hats, and they might want something purpose-built out of the box, and so when we built our, you know, internal harness, if you will, we started then seeing, okay, what are… what are folks using? What do folks like about this? How can we recreate you know, the best parts of the context graph that we built internally. How do we make sure entity resolution, identity resolution, it’s very easy for us to do internally. how can we then go build something that folks can use externally that might not have access to everything that we have access to?

And so, you know, we do a side-by-side, and then I think to the earlier point, you know, that judgment piece. I think right now, as a seller, that judgment piece is more important than ever, because I think we’re all in some form of build versus buy, if you have consumption-based pricing versus traditional SaaS pricing. I think one of the interesting bits of noise that I… it’s interesting call by call as we go through.

But… this concept that everybody wants to go on consumption pricing after they’ve just been token maxed and they ran up a, you know, million dollar bill with a… then they’re like, I never want to pay consumption pricing again, I want fixed pricing that looks a lot like legacy SaaS pricing, and then you have some customers that are like, I’ll never pay a fixed seat cost again. And so, if you’re a rep, not only do you have to figure out, okay, build versus buy, am I going headless, am I going more traditional, out of the box? everything in between, and then how do I price this?

And, you know, we have agents that will say, okay, Zoom video is on this journey, this is absolutely the conversation that you should be having. you might not have the fact that they just got that million dollar bill from someone else, and so, you know, you have to make the judgment on how do you price, what, you know, product route do you go, headless versus out of the box, and so, you know, I think taking all of those learning lessons internally and then building them into Ultimately, what matters most, which is, like, if we’re customer zero. what are our folks getting the most value out of?

How do… you know, if I were to take that product away, what are the things that the entire floor would be screaming about? And then building products that we can then get to our customers that get them, you know, 90% of the way there without having to buy millions of dollars worth of, you know, clawed or otherwise, so…

Seth Marrs:

So, classic positioning, right? What do you need, and let me get it to you in the way that makes the most sense for you and your business.

James Roth:

Yeah, and I think with that context, similar to verticals, you can get a lot closer to the pin on where that customer actually is. I think we’ve always said, be the trusted advisor, meet the customer where they are, but then here’s the boilerplate pricing, here’s the product you’re gonna jam down their throat. Now, there’s more flexibility in product and pricing offerings, and you have so much more context in terms of getting pretty close to what that customer is probably going to want, and then you leverage the human and the judgment and some of the art, if you will, that Stevie spoke to.

I think to ideally get to the best outcome.

Seth Marrs:

So I’ve got to ask this question, just because I know it’s on people’s mind, even though it’s a little bit off-topic. Is there gonna be a way for any company to avoid charging consumption, now that every company on Earth has a foundation model tax that they have to pay? Is… like, I mean, anybody can talk about it, because you guys are all in this world. Is there… is this the end, like, will… I get there’s going to be hybrid, but how do you overcome that tax if you’re a company, if you want to survive?

Graeme Geddes:

Yeah, I would just say that, you know, what we see is very similar to what James had mentioned, right? So customers, we see a huge cohort that are saying. look, I don’t want the unpredictability of the consumptive model, and they’re actually reverting back to, right, it might be in vogue again, the traditional, you know, SaaS, you know, seat-based pricing. Many are leaning forward in hybrid, which is, you know, the seat plus consumption. And I think what we’re seeing is, you know, as with any technology, you know, as we start to see it democratize and look we’re seeing year-over-year costs of intelligence, you know, dropping dramatically.

I think that there could be a world over time where, you know. kind of the pendulum swings, and it lands somewhere back in the middle, where we can lean in, and giving customers what they’re looking for, right, which is the predictability, but also with extremely powerful technology. Because I… right now, what we’re already seeing is that a more, you know, if we move from You know, 5.5 to 5’6 on a model, that doesn’t necessarily change the outcome. Right? It’s the contextual understanding, it’s the graph, right? It’s the harness that’s even more important. And so, if we take that as an assumptive. then what I think we will see, right, is the ability for us to provide more packaged solutions, right, in a predictable model for our customers.

Seth Marrs:

Fantastic. Yeah, because right now it feels inevitable, at least in the next year or so, because… and you’re watching companies kind of stretch to it, so it’s… yeah, it’s really interesting.

Graeme Geddes:

Yeah, but I think it goes back to what you had said before. It’s meeting customers where they’re at. Some want 100% consumptive, will lean in. Some want 100% seat. The other one, right, that’s very popular is outcome. based pricing, where I don’t want to pay for the consumption or a seat, I just want you to do a given task, and that’s an area where we’re leaning in heavily as well.

Seth Marrs:

Got it, got it. Very interesting. Thanks, thanks for… for… let me go a little off-topic for a little bit, because I think the audience is… is… is… if you’re not dealing with that, I’m sure you will be. Well, so, one of the things we talked about was next best action versus next best option. So, I want to get your guys’ perspective on it. Which one is the future for sales? I think I have an idea, but I’d be interested in your guys’ perspective. Niroshan, do you want to kick us off?

Niroshan Rajadurai:

I mean, again, from our perspective, next best action versus next best option, I mean, we’re obviously building out the motion here, so next best option for us is still very early in that piece. Next best action is something that we have a lot of data on as we’re building out pipeline, building out all these opportunities, capturing the DNA of, like, you know, what does a good motion look like with a customer, and then very quickly trying to replicate that across the organization. So, for us right now, it’s about action, it’s about getting lands, getting engagements with customers.

But I think over the next, like, 18 months, as these lands start expanding, like. determining what’s the next best option for the customer, what’s the, you know, where do they want to go, we’ll start to have better data. And I think part of that also is just the conversation you had before around, like, procurement models, right? So, one of the things that we see today is a lot of the customers that we talk to have very big token, or very big inference commits with the Frontier Labs. And so the challenge is, like, you know, what’s the next best option for a customer?

Well, you know, depending on where their commit is today. how do they want to procure? They may be lagging in that other area, like, they may be lagging inference burndown. or in their commit to the Frontier Lab, so they want to procure in a different way. So I think, like, that information, as we build out our go-to-market motion, like, next best option will start to fall out more for us. But right now, we’re just very focused on, like, how do we actually get in front of a customer? How do we land that customer and start proving out what is, like, brand new technology?

And driving that path forward. So that’s where our focus is right now. So next best action for us is the short-term goal.

Seth Marrs:

So, like, just to expand a little bit, like, action being, I give you one option, this is what I want you to do, seller. Is that where you’re focused… so it sounds like I’m focusing on that, because you’re building your modeling around what works best. Do you see yourself extrapolating where you allow the seller to pick from? Or you give them data on… there’s 4 options that have worked, and here’s the relative, like, value of each one, which ones have worked at what percentage? and then track that back? Or do you see it as, I’m just gonna give you the best action, and I need you to either follow it, or not follow it, or… Like, how do you…

Niroshan Rajadurai:

I mean, in very early stage, it’s better to constrain the options, otherwise you have an unmanageable surface. So, from our perspective, it’s like, you know, based on the signals that we see, we want to recommend, like, here is the thing that we think you could do. Otherwise, maybe right now, this customer’s not the right fit for us. As soon as you start, like, right now, if we have, like, put multiple options in front of the seller, like, it’s just unscalable, at the size that we are. But I think over time, you know, definitely we’ll be in a position to put those different things there.

Seth Marrs:

Now, Philip, you’re in a… I think this may not be the case, I think you may be in a similar case with Niroshan around your prospecting side, but it sounds like you have significant richness of data around your current customer base? Yep. How do you see… how do you see that from that perspective? Like, the option versus, I’m gonna give you this and I need you to do it?

Philip Lacor:

Yeah, look, It’s the next best option for us, because at the end of the day, these systems are probabilistic and not deterministic, so… There’s still, like, ambiguity around, like, customer situations, and although they’re… Although you have a lot of data and a lot of signal around your customers, there’s still… it’s still a probability. And, so that’s… that’s point number one. You don’t have absolute certainty. Number two, we also want, like, critical thinking from our reps, right? So if you want judgment. Then you want to stimulate that they keep thinking about what is the best thing to apply.

And, we have for our existing business, for our account managers and CSMs, we have this thing called Copilot, and it will literally show, when you’re looking at a customer, red, yellow, green, here are the things that you can do. But, I compare it to a Kady when playing golf, right? So when you see a Kady and a professional golfer, they debate, they argue before they hit the ball, and they disagree on, like, which club to take and where the wind is coming from, and in a way, we want, If you want to fuel that critical thinking from your reps, then, yeah, you should give them options and not say, hey, this is the one and only answer.

Seth Marrs:

Got it. Niroshan, kind of going back to you a little bit, I mean, is some of this just a magnitude of demand on the front end, and you trying to go pace, so I need you to act quickly and Like, Philip, like, talking to you about it, there’s, like, the deal… complexity level is… is… you’ve got a customer, you know for a long time, that there’s a lot around understanding that. And Niroshan, you’re growing so fast, it seems like… There may be a case for both, depending on where you are in your maturity cycle.

Niroshan Rajadurai:

Yeah, so I mean, we always listen to humans over AI, like, I mean, that’s rule number one. But I think at the same time, you know, our goal is really, right now, is to scale very quickly. And so, there’s a golden path that we think that, you know, we should be following. It’s always possible for a rep to look at that and say, hey, you know, or I need to look at that and say, you know, in my experience, I have a different opinion, I have this other things, but we’re trying to avoid, like, over-complicating the decision process in terms of what we recommend.

Because from our perspective, if we can constrain a recommended path, then we can keep them on that… on that golden trail, and then the rep can… or the AE can always come back and say, actually. I have another option, or I have another perspective, and we can always have a conversation around that, but from our perspective right now, it’s… we really want to be focused on where… where we think we can win and where we can be successful, and then take human feedback as part of that loop.

Seth Marrs:

Got it.

Graeme Geddes:

Except maybe I’ll just comment really quick, just, you know, the next best action, you know, kind of idea is something that’s actually existed for a long time within the customer support area. So when you call in, you know, a customer support rep, there’s been technologies to provide next best action for them, you know, whether it be, you know, kind of initiating a return or what have you. And in those systems, it’s not… an OR? Right? It’s an AND, so it’s next best action AND optionality, and so the AI can actually give you a predictive score to say, hey, with 100% certainty, right, here’s the refund that we should do, or guess what, I’m unsure, and 80%, but here are the other two options.

And I think what we’re seeing is that same approach. really, really working from a sales perspective, so we’re building these technologies into our Zoom Revenue Accelerator with Live Assist, so we can give that next best action to the rep, but what we’re going to give you is kind of that scoring to say, hey, in this situation, I’m… you know, partially certain, and here are the other two options, but I think the score is the important part, because then it does put it back, you know, into the value of the human judgment, to actually make the determination of which action do you want to take.

Stevie Case:

This is one place where we’re seeing an interesting bifurcation, because at the scale that Vanta operates at, it’s 800 people, we are trying to adopt a lot of that customer service, customer success mindset, and if we’ve got paths with one certain next action that is recommended, we’re trying to automate that wherever possible, and that’s not possible everywhere. You know, certainly as we serve our more complex customers, there’s a lot more judgment required. And it makes more sense to have options and a human apply judgment, but where we can, we are trying to map the universe of next best action that is deterministic and can be set to a single action, and that is exactly what we’re automating.

So we’re trying to carve that off for the bots, and then everything else that requires judgment is where we bring in humans.

Seth Marrs:

Interesting, so if it is an X-best action, you might as well just have an agent do it.

Stevie Case:

Absolutely. If you know with certainty, with a high… at least high enough confidence that there is one next action that is best, then that is something that an agent should be able to do.

Seth Marrs:

Great insight. Okay, I have one final question for you, for you guys. Let’s flip this around. Are you… are any of you starting to see Buyers using similar tools to give you a next best action, or to give you responses and take you through a path, and using what would be adaptive buying tools to counter all of this stuff you’re doing to try to help position yourself to win.

Graeme Geddes:

So, maybe I’ll start here. So, I think we are starting to see buyer personas be, you know, showing up more informed, so they’re very much leveraging AI to understand and, you know, kind of inform themselves across the traditional buyer journey, which I think puts the burden back onto the seller, right, to really, you know, understand, you know, the elements that, you know, kind of the human you know, can drive through the sales engagement and sales process, right? Discerning, right, who is it that’s ultimately going to make the decision? How do I make sure that I’m influencing them?

So that… but I could envision a future point where, you know, the customer bot is talking to the, you know, the vendor bot, right? And those, you know, kind of do some of the blocking and tackling up until the point where they’ve come to an agreement and it escalates into kind of a human interaction. So, I think it’s very, very early here, but we are seeing, you know, buyer personas showing up leveraging AI and Agentic systems, right, in that kind of initial, kind of, I would say, kind of the early, early parts of the phase, and it’s offloaded a lot of the sales process as well.

So it’s not just a, it’s not just a negative, right? It’s actually a positive, where there’s a big part of the sales, kind of, transfer of knowledge, or, you know, helping understand their… their careabouts that are done, kind of, before the rep is involved.

Seth Marrs:

Got it. Got it.

James Roth:

I think just to add to that, you know. anyone who’s been selling stuff for the last couple years, there was this emergence of, kind of, outsourced procurement that would work alongside of procurement teams. There was a couple big names, there was Vendor, Tropicana, and, you know, they came in and did exactly what Graeme’s talking about. They had benchmarks, they had pricing, they would… at least sell the fact that they knew how to get certain better pricing from any of us. And I think that’s an area that we’re seeing a lot more of those sidecars, where it’s not that there is the procurement bot that is just, you know, asking us for final deal points.

But it is more of an Agentic version of what those companies did in the past, where they’ll bring all that data in, they’ll have their, you know, customers, they’ll look at Reddit. And so I agree, it’s just a far more informed buyer, but I… we have yet to see the agents doing deals, on both sides, and so I think it will continue to move that way, and I think from a… I think the area that you’ll probably start to see it probably more than up in the enterprise is, you know, any kind of PLG-led or low human touch There’s probably going to be a low human touch procurement side of that as well, and I think we’re probably very early innings on the consumption side.

I imagine that’s another area where you’re going to start to see Agentic procurement come in. You know, you have the governors from the product that you sold to the company that tells them how many credits they’ve used, or how many tokens they’ve used. I imagine they’re gonna build out, you know, the ability to have a kind of a procurement cop that sits on top of consumption internally that then comes and yells at whoever the provider is, so… I see it same as Graem, where it’s early, and it’s clearly helping the procurement side, but not necessarily, you know, across the board, full Agentic.

Seth Marrs:

Yeah.

Philip Lacor:

I think where it is important is in the top of funnel. Obviously, search is down, LLMs are up, so you gotta be on your game when it comes to top of funnel.

Seth Marrs:

Great point. Great point. I know we’re up on time. Thank you guys very much. I learned a ton. I guarantee our audience did, too. Really appreciate all of your insights. Julie, I’ll hand it back to you.