20Sales: How the Best Sales Teams Use AI to Win Enterprise Deals | Sales Teams Will Be Dramatically Smaller | How to Ramps Sales Reps Way Faster | Why Unpaid Design Partners are BS | Why this Generation of Sales is Soft with Ishan Mukherjee @ Rox

Will sales teams be dramatically smaller in the future?

Yes, dramatically smaller. It could be 10 to 20% of current organization sizes. Sales is a power law. The top 10% bring in 90% of the revenue.

Do you think we have a generation of salespeople who are not ready for the challenging nature of sales that is to come?

Yes. We had it relatively easy the past 4 to 5 years. I don't want to be morbid, but things are hard. Winning is hard. It's hard, hard work.

Ishan, I am so excited for this. It is so much nicer to do this in person, so thank you for joining me.

Thanks for having me on, and it's great that we put it together on such short notice.

It is lovely to do. I want to start with a little bit of the foundational story. New Relic from zero to 100 million—I want to start with the PLG motion that you—

Okay.

…developed there. There are so many founders listening who are building their PLG motions as we speak. What did you do specifically well that you think was so instrumental to the growth from zero to 100 million?

The central word is focus. We centralized the entire go-to-market engine on a captive audience of developers for whom New Relic was still the default brand. New Relic had lost its aura, but there was an audience of about 100,000-plus developers who joined the workforce from 2008 to 2014. For them, New Relic was still the hot brand.

We shut everything down, reallocated investments to go after them, and earn the right for them to use the product again. That's what the whole machine was built on: divest everything else and invest where we are the default. We came into a position where the company had reduced its investment in or shut down the PLG motion, so there was this latent demand that we just had to unlock.

We went back out to the developer community and said, “Hey, we have an unlimited free offering. You get all of what you need in a pretty cost-effective manner.” That was like drilling into an oil reserve.

Do you think it is possible for early-stage companies to do PLG and enterprise at the same time?

It's very hard.

Mm.

It's only possible if you have a dialed-in CEO or C-suite who really gets it. If not, do not do it. You need a singular person who understands the motion. If not, I wouldn't do it.

What do you mean by a C-suite that gets it, and what does investing in PLG actually look like in reality?

Absolutely. The core idea is that you can invest in bits or atoms. Bits are the product; atoms are distribution or sales. A person who comes to mind is Amit Agarwal at Datadog. Not a lot of people know about him, but he's the brains behind the product-led distribution engine at Datadog.

The core idea is: how can you keep investing in the product so that it does most of the product sell, and then layer in humans to do the commercial sell? You have to have a singular figure who understands engineering, product, sales, services, and support to really manage that funnel on a day-in, day-out basis. If not, it's going to become 2 or 3 siloed organizations that just don't work effectively.

Given the rise of PLG and the centrality of the product being so good, with people wanting to touch and feel it so much sooner, and the importance of marketing and content, as we discussed before, does the role of sales become ever more reduced in a world where PLG starts, marketing helps it along, and then sales is signing the enterprise deal?

Ultimately, I do think the largest customers need humans to actually work with them to build non-transactional, long-term relationships between a buyer and a vendor. What I look for is customer pull. When the customer is pulling the company along, both in terms of building out the roadmap or coming up with a commercial construct, that's where you have to layer in enterprise sales.

I fully believe that you need humans because customers want their problem to be solved. They don't give a fuck about products or services. They want their pain to be solved, and they want to build a non-transactional relationship with these vendors. So that's where sales comes in.

My view is: get the product to do product sell, and get humans who are super strategic to do the commercial sell. But they're doing non-transactional commercial work. They're not just signing an order form. They're really working with the customers, figuring out the multi-year relationship and a roadmap, both from a product perspective. That's how I think about the future of sales: the top 10% who are working with customers and building long-term, non-transactional relationships, both from a product and a commercial standpoint.

So, we unlock this PLG motion—

Right.

Okay, that was latent demand. What did we do specifically well? You said there was focus. Specifically in terms of the investments that worked in PLG and those that didn't, what are the elements that come to mind when you unpack and review what worked and what didn't, and how that shapes how you think?

Absolutely. I am not a seasoned go-to-market executive. I'm a first-principles founder. This is what I live for. When I came in, I just had a simple Jason Lemkin funnel mindset: let's break out all the channels and break it down into awareness, consideration, and conversion. We're going to break down the funnel.

When you look at it, most growth or marketing organizations will be stretched too thin. They'll invest in six or different channels at the top of the funnel. Tactically, what it meant for me was that, if you look at the data, organic content and performance marketing worked really well for New Relic. New Relic was the original brand that did land and expand.

Everything else—events, DevRel, and other pieces—you do because people want you to do it, but they don't really work. I went out of my way and completely shut down event marketing. We don't need to go to conferences where nobody comes to our booth. Why? Not because we couldn't do a good job, but because we couldn't be world-class at it.

Datadog and Ahmed were pouring all their capital into re:Invent, and they were doing a phenomenal job. What's the point of spending money going to re:Invent when you're a medium-sized booth and your competitor has the biggest booth? So let's just not do it. Play to your strength and dominate that.

I shut everything down and focused on performance marketing, which is SEM and SEO, and then organic content. It's a slower boil, but just do those 2 things.

The rest is distraction. Then you get to the next phase, which is consideration. You’ve got the eyeballs; how do you get developers to consider your product? The biggest mistake I made—I made so many mistakes, more mistakes than successes—is that people treat a free-tier product as a converted customer.

They’re not converted customers. A free-tier product is just a better website. People come in; instead of browsing the website and marketing, they’re tinkering with the product. So just move your free-tier customers to, “Hey, this is consideration.”

What that means is that now we’re investing in converting customers from consideration to converted customers in the product. Those were the 2 singular things. Once you do that, you don’t have resources, so you just start divesting everything else.

I get worried when I see startups doing a lot of booths at events. You don’t have the money, as you said. Would you say that events don’t make sense for the majority of startups?

Yeah. I’m somebody who goes all in on things that work, and then you bet on finding the next thing. If events aren’t working for you, if you don’t have a dialed-in operation, don’t do it.

I think you’ve also got to take a really holistic approach to events, which Ramp does really well. They’ll sponsor the key card for your hotel, sponsor the billboard outside, and do 5 other things that make it a really multichannel effort, not just a booth.

Absolutely. One of my ex-colleagues and now a mentor, Manav, who’s the chief product officer—he was early at Twilio, early—

We spoke to him.

Yeah.

Yeah.

He’s amazing. He’s probably the greatest executive I’ve worked with. I worked at Amazon and Apple—he’s that good. These are the operators who build businesses from the ground up, who don’t look for the fame and glory but are really, really good. His point was that the dinner after the event, around the conference, is a lot better than the conference itself.

Yeah.

Right? The best thing to do is take your dollars. We still did the events; you just don’t do the booth, and you don’t do one of 100 sponsors. Just have an amazing baller dinner—

Done.

Right after. Have an authentic conversation. You get 2 or 3 deep relationships.

Moving a step along, how do you think about SEM and SEO? That will shift so significantly over the next few years, bluntly, with the rise and dominance of ChatGPT. I’ve also seen Guillermo Rauch at Vercel mention that it’s moved from 1% to 4% of their inbound. How does that change, and how does that impact how you think about it?

Mm-hmm.

I think SEO is a long-term strategy. It takes a lot of time to build credibility, and that will change in terms of form factors, with AI answers and search—Perplexity-like systems, I guess—but SEO isn’t going away. You need a core engine that produces high-quality content that’s indexed and discovered in new form factors.

For me, it wasn’t an option. I owned a 14-year-old brand. The first few blogs still outworked every new blog that you wrote. What I look for is the top 10% to 15% of blogs or pages that return 90% of your value, and then distribute them really well. So I think with SEO, you still have to produce super-thoughtful content, but just do fewer, higher-quality pieces and make sure they’re distributed.

With SEM, you want to be proactive. I came at it from first principles. I think it’s a complete cartel. If you really look at how you invest capital into the SEM game, you’re just selling your soul to a pretty messy market.

Sounds like venture capital.

You’re bidding on keywords, topics, and subject matter. If there’s money to be made, each keyword or brand name is a market of its own, with competitors bidding. So it’s really hard to figure out the signal-to-noise ratio.

But for us at New Relic, we’d figured out there were 10 to 20 keywords where we were really good, and we could put customers in a stitched path. They could see an ad, get to a landing page, use the product, and get to an aha moment. These could be very boring things: “How do I debug my Kafka key with New Relic?” Very specific.

Those made a lot of sense. We could pour money in and see returns within the quarter, maybe within the month. The rest, I really don’t buy that there’s ROI, and that’s where SEM is headed. If you can get really good at capturing intent and producing a stitched path, keep investing. On the rest, I’m very skeptical.

I think this is the thing that everyone forgets about content: it often takes 2 years before you even see any ROI come back.

Exactly.

And so patience is everything, so I totally get you there. You mentioned consideration: if it’s free, it’s like a website. I see a lot of people today doing design partnerships and heavy collaboration with early partners instead of PLG—letting everyone test, and then iterating from there. How do you feel about free design partners and building with them versus staying open, iterating, and learning?

We don’t do any free design partners. We really focus on paid customers with a pilot clause. If the customer isn’t parting with some sort of cash, there’s just not enough skin in the game for them to put in the time to work with you for a few weeks, a few months, or a few quarters to build a product.

I’m very anti-unpaid design partners because the cost for them to just say yes is a lot lower than it is for you to keep making them happy. So my view is: first have paid design partners who are on an alpha or a beta. We’re on an open beta now.

In parallel, the product has to be open, especially if you’re competing with incumbents. Incumbent enterprise software is gated by demos and implementation steps. It’s a 3- to 6- to 9-month time to value. If you’re a product-shipping machine, you have to have it out, even if you’re extremely embarrassed, like we are at Rocks.

This isn’t perfect by any means, but you’ve got to have it out there because, at the very least, it reduces friction from a sales motion. At the core is this undeniable confidence that you can ship your way out of holes and also ship your way to the Promised Land because you’re not hiding anything. So my view is paid pilots and paid design partners, and then, as the product gets better, keep it open. Don’t put gates.

You mentioned incumbents. If we go back to New Relic, Datadog bluntly dominated the market over the last few years.

Absolutely.

Why do you think they won in a way that New Relic didn’t?

Initially, I thought it was product. When Zan, Eric at Benchmark, and I were starting Pixie, we thought, “We’re just going to build a 10-times-better product.” But when you actually compete with them day in and day out, you realize it was consistency—specifically, consistency in go-to-market execution.

You’re in the trenches, competing with Datadog, Splunk, and these other established public companies. You realize that Datadog is a machine. Olivier, Amit, and their whole crew never changed segmentation or comp models. They just ran a dialed-in machine for 6 to 7 years, and they compounded value while everybody else kept thrashing: changing pricing models, changing segmentation, and changing comp. It seems very boring at the surface level, but you’re just—

I thought we were always told to be reactive. “Hey, you’re always going to change your pricing. Be flexible. Be plastic-minded.” I thought we were meant to be reactive.

Yes and no. The no is that, including myself, a lot of folks want to reinvent the wheel with enterprise sales and distribution. There are certain things where you don’t want to reinvent the wheel, like how you segment customers and the comp model, because the cost of reinventing the wheel is that you’re avoiding the ability to compound.

For example, at Datadog and a bunch of those companies, if you look at their enterprise reps today, they joined as inside-sales reps. They’ve grown through the ranks. If the segmentation and comp were changing, you just wouldn’t have this talent flow of people who know the product, know the customer, have seen everything, and are now at an elite level. If you’re changing it all the time, you’re just not compounding any institutional knowledge.

You mentioned reinventing the wheel where maybe you don’t need to or don’t have to. In a world of AI, do we need to reinvent the way that we do enterprise sales?

100%, across the stack. And that’s what I learned the hard way. You talked about New Relic. Obviously, we had a lot of success in the first part of my time there. In the second half, on the backside of COVID, we realized the game had changed.

Specifically, the way you sell and the revenue models changed, from commit-based revenue to usage- or consumption-based revenue. Now the dollar values on the contract were just words. You only made money when people actually used the product.

When you do that, you have to have the right leader, retrain your entire frontline workforce, and fundamentally change the underlying systems. That’s where Rox comes from. That’s the core thing: that has changed. And now with AI, what we’re seeing with Cursor and Cognition on the building side is that the best are going to get better.

Your President’s Club sellers, your best VP of Sales, your best CEOs like Eric at Ramp—they’re just going to get better. They’re not going to stop. This is an information-arbitrage game, and they’ll get better. So the rest has to adopt AI to fundamentally get better, or they risk being irrelevant. Sales is a power law: the top 10% bring in 90% of the revenue, and it’s going to get more acute if you’re not using what is now—LLMs are the new computer.

There’s so much for me to unpack there. With the change from commit-based to usage-based contracting, how does that change how we need to invest in our different functions? Do we suddenly have to have a huge amount more in customer success because we fundamentally need to make them a champion as soon as possible?

There are 2 schools of thought here. There’s the Frank Slootman school of thought and the non-Slootman school of thought.

I’ve had Dagnan on the show from Snowflake. He’s like, “Fuck customer success.”

Exactly. The Slootman school of thought is: only invest in the bag carriers. Get the AEs, the people who can hold a number. You eat what you hunt. Those folks should hold both the commit number and the consumption number. It’s an intense way to operate, but I actually think that’s the right way.

Is that even possible, though? If you’re continuously chasing mega-deals with your biggest companies in the world, and then I’ve also got to build internal champions, get wall-to-wall adoption, build multiple champions, and multi-thread, I’m one person.

Exactly. What I’m seeing is that now you have AEs, AMs, and sometimes SEs, or sales engineers. They’re all—

What are AMs?

Account managers.

Ah.

If you’re a large account, you have this relationship-focused but strategic account manager.

Which is CS, no?

CS was traditionally designed as a function to onboard the customer and make sure you go through the checklist. Customer satisfaction, or CSAT, was their primary metric. Expansion or renewal wasn’t their primary metric. AMs completely hold a number for expansion and renewals. Their job is to manage a large account to make sure there’s growth across the organization or depth within the existing teams, and they’re comped to a certain degree on the upsell.

Again, the philosophical thing I learned the hard way is that not every human can hold a number.

How did you learn that?

By making mistakes.

But what showed you that, though? What was the lesson that taught you that?

Again, like you and a bunch of founders, the best way to get it is you have to go close to the source. I just join customer calls. I own and manage customers. You can tell how people behave.

What does it mean, though, to not own a number? You just can’t hit the number? What does that mean?

There are 2 kinds of salaries—2 ways people make money in go-to-market. You might have, let’s call it, a 50/50 split: 50% of your compensation is salary, and the other 50% you only make if you can bring in revenue—commission. The other is that 100% is fixed, and then you could get a bonus.

Those are 2 very different things. The first one is traditional for salespeople. The term “eat what you hunt” means you only get your full salary if you bring in the revenue. That is holding a number: I have to bring in revenue to actually put, in some way, food on the table. That’s how a CEO thinks about it. That’s how a founder thinks. It’s this owner’s mindset.

I do believe that it’s a hard path, but that is where modern businesses will go. You will invest in people who have full skin in the game and have this owner mindset.

You said that Frank Slootman was one school of thought. What was the other school of thought?

The other school of thought was the more traditional model, where every account has account executives, AMs, CS, customer success, and sales engineers. It’s a 5- to 6-person pod, but there’s only 1 true hunter there. The AE is the one who’s actually commissioned. Everybody else is on a fixed salary.

Yeah.

That’s a fine way to go about it. Datadog has a similar model if you have really good leadership and can manage the incentives well. But in most companies, like in my case, it’s really hard. As Charlie Munger says, “It’s all about the incentives.”

It’s really hard to set up the incentives across a 5- to 6-person team to make sure they’re all focused on making the customer happy and actually increasing usage or consumption.

Dude, you’re a pro.

Yeah.

What are your biggest lessons in sales compensation? Imagine you’re a first-time founder now with Rox, and you’ve never done this before. It’s really hard to know what to do. What would be your advice to someone building sales compensation plans for the first time, knowing all you know?

Now I support some of the best go-to-market leaders in the world, and honestly, I know 0.1%. If it’s the GC at OpenAI, or these folks, they’re legends.

What I see is 2 different approaches. Some CEOs, CROs, and CFOs want to run a go-to-market engine where they’re really going for high revenue targets, and they want to keep the vibes good. They want more and more people to be attaining 80% to 90% of quota.

The others really want to see a normal or bimodal distribution, with the top 10% at the top. So there are actually 2 different philosophies. The first philosophy is that you set up the compensation in a way where everybody’s winning. Yes, you have your club killers, but most people are hitting quota. That model works if the whole company’s growing, the TAM is large, and you’re getting there.

The alternative is the power law, which is now getting more and more extreme. My general view—I’m doing early-stage sales at this point—is that, in early-stage sales, the most important thing is to find these outlier sellers who can sell without a massive brand, a demand-gen engine, and perfect product-market fit. You’re trying to find those outlier humans.

Do you think, as we move forward into the next decade of sales, that we’re finding these outlier humans and making them even more exceptional? Does that not create a 2-tiered system, where you have this 1%—suddenly the 10% become the 1%—and they leave behind all the other people who maybe just aren’t as good, bluntly?

It’s hard to avoid the impacts of technological progress.

But it’s hard when it’s in the same team.

Right.

Do you see what I mean? It’s fine if Datadog’s exceptional and then there are shitty people elsewhere. But if there are killers in Datadog and I’m really average, it creates this 2-tiered culture.

Absolutely, and that’s where management and leadership come in. The business owners and the leaders have to understand that the distribution’s getting more and more bimodal. You have people who are falling behind and people who are actually rising.

The opportunity is to completely retrain the workforce. The job has changed. The way you do your work has changed, and you hope to have them evolve and thrive in the modern economy. If not, it’s an existential decision.

So, on sales, as specifically as possible, how has the job changed in the way that a salesperson works?

There are 2 phases as a salesperson. First, you ramp to the job. You learn how to actually do the job, and then, while you’re doing the job, you’re trying to hit or exceed your targets.

Right.

In the ramp phase, what used to take 2, maybe 2.5 quarters to ramp a rep—you’re reading everything about the customer and the product and going through training and so on.

If you’re a really savvy person, you go to Perplexity, you go to ChatGPT, hopefully with Rocks, and you understand everything about the customer and everything about what you’re selling. You can really ramp much faster. So you’re talking about gaining a quarter to a quarter and a half.

How do you ramp faster? What is it that allows you to ramp faster?

It’s deep research, understanding the pain, and building your own internal frameworks for how to communicate your solution to that pain. At the nitty-gritty level, the game of sales hasn’t changed for a long, long time.

What people used to do was print out 10-Ks, 10-Qs, and SEC filings, and reread every CEO transcript. Now you can do that on your phone in a few days. It’s just reducing the time to do research and forming your own opinion.

Do you think people know what research they need to do to get that information quickly?

The best do. Rocks focuses a lot on enduring enterprises. We work with large, legacy, old-school enterprises. When you spend time with 20- to 30-year vets who’ve been selling into automotive or airlines in Europe, they get it because that’s the only way to earn the credibility to even get a conversation, let alone move pieces.

What has happened over the last 5 to 7 years is that people think this is easy. It’s not easy. You have to be a consultative expert in the customer’s business, and you have to do the research, just like you all do on Founders. That’s the only way.

We call Manav beforehand and ask tons of questions.

Yeah.

I totally get you. Do you think we have a generation of salespeople who are not ready for the challenging nature of sales that's to come?

Yes. And it applies to most functions in venture. Founders and builders had it relatively easy the past 4 to 5 years. I don't want to be morbid, but things are hard. Winning is hard. It's hard work.

If you look at sales specifically, when everybody's spending and everybody's buying, it became pretty easy. You're becoming order takers, as they say—shooting fish in a barrel. But when you're selling something new or selling something large, there's no easy way around deeply understanding the customer's pain and how to actually solve it.

I'm seeing that constantly. Honestly, candidly, we've done really well over the last year, but it's taken me maybe 2 to 3 months longer to find those outlier sales reps because it was pretty easy.

It's like you didn't need the outlier sales reps because people were spending so freely and sales cycles were so compressed.

Yeah.

I totally get that.

Yeah.

There are so many areas that I want to dig into. You said that Phase 1 is the ramp and the compression of ramp from 2.5 to maybe 1.5. If I'm a sales leader today listening to this, what can I do to reduce the ramp of my new reps?

Give them the power of data and tools, and reduce friction.

What does that mean? That sounds wonderful, but—

Exactly. Very basic things. Most sales reps in the world use ChatGPT, Perplexity, and LLMs in their private lives, but they can't use them in their professional lives.

Because of data access problems.

Data access. These seem like simple things. I think the world has changed. Just give them access to the tools.

Second is, when you think about how you train and onboard people, it's very similar to schools versus teaching your kids through YouTube. You have to completely change how you train people by making it more AI-enabled, so that you basically get them in front of deals and give them live customers, versus having them go through 2 to 3 months of old-school training because that information is stale.

How do you think about how quickly to put reps in front of customers? Customers are precious.

Yeah.

You must treat them so carefully.

Yeah.

If you put a rep who's not ready in front of them, it may not be the best representation of your company.

Yeah.

How would you advise me?

This is where I'm not the expert for scaled go-to-market. Even at the early stage, my mindset is still the same: get them in front of customers with mentorship as fast as possible.

What I do at the early stage is, after a few weeks, you have to shadow a ramped AE or myself. Without being in that room, it's just really hard to replicate what's happening in the world. The world is moving too fast.

I'll give you an example. We have a capability where you can basically do deep research on your customers. 4 weeks back, people were wowed by what we had. 3 weeks later, Deep Research launched. They were like, “Oh, no. I now expect PhD-level research.”

If your training doesn't keep up to date, if you're not on the front line, you just won't see customer expectations change because they're changing every day. My general mindset is: hire somebody. In the 1st week, I have a sense, based on the questions that they're asking, of whether they have it or not.

What are those questions that make you reflect and think they're great candidates versus not?

There are questions that they ask me or us, and then there are questions I ask of them. If they're asking us questions to deeply understand the psychology of the customer and the buyer—things like, “Hey, what is their expectation? What do they think of you?”—those are really good questions.

The thing that you have to really understand is: what is their pain? What is their propensity to buy or sell? But now, deeply, every consumer's expectation of software has changed. Where are they in their journey? Do they get it? Are they early? Those questions are important versus any boring discovery question, right?

Are consumer expectations high when you're selling into the biggest enterprises, as you do, when they're traditionally working off pretty shitty legacy enterprise software in their professional lives?

Right. This is the first time I'm seeing what they're doing in their personal lives carry over to their professional lives. In their personal lives, they're teaching their kids with ChatGPT, Anthropic, and now they're coming in with a totally different expectation.

If you look at the intent logs of how people are using Rocks, they expect software to not just do some work. In some cases, they want it to do all the work. Humans across the board, including my mom in India, have obviously been sold on the promise of AI and LLMs.

When they come into the workforce, they actually want the software to do as much work as humanly possible. The specific word is “humanly.” The software should do anything that a human can at scale, so that they can do more—and more, in my domain, is supporting more customers.

Why does it make sense for Rocks to be verticalized versus people just using ChatGPT?

It's data. The alpha is in being able to index all context about your customer.

Oh, so I give you my CRM?

Exactly. So Rocks runs in parallel to—

So how does this work?

How does this work? Very good question.

So I give you the CRM, and then you get all of the company-level data of all of my targets and all of my pipeline. What does that allow you to do?

Let me go back to the origin story a little bit because you brought up New Relic earlier. When I was competing head-on with 6 or 7 public companies, the difference I was seeing in the trenches, in customer calls, was that most account and customer-facing reps were just not proactive enough.

They knew a partial view of the customers, but they had no real-time understanding of: How much are they using? What do the support tickets look like? What are they looking to buy? How are they growing? Not everything was in the CRM.

The future of sales is figuring out how to feed in real-time context—hopefully, all complete context about the customer—so that the human in front of the conversation is essentially supercharged with this kind of Batsuit and is actually moving the deal along.

What we do is, Rocks gets installed in parallel to your CRM—HubSpot, Salesforce, or Dynamics. It runs in your warehouse. It indexes all internal context and brings in public context, and it's trying to do the work for you.

You have a team of Rocks agents, or this team of analysts, working for you, so that you have full context and are basically moving the conversation along.

Now I get it. That was fantastic. I'm so glad. Totally makes sense. What is the primary task that Rocks does today for sellers?

Absolutely. Today, it does all customer research and prioritization. We're doubling, tripling the customer's books, and now they have tens, sometimes hundreds, of customers they're dealing with, so it does all the research and prioritization for them.

So now I can outbound way more effectively with way more personalization.

Absolutely. The second is outbounding. We're the anti-AI SDR. My view on automation is that the human evolves to become an orchestrator, a quarterback.

We do all the work of composing your sequences and deciding who to reach out to, but the human reviews and sends. This year, we look at this data pretty closely. We're not at a point where a CEO or a CRO can just fire in the hole and fill the pipe with these outbound systems because it's going to dilute your brand.

On the outbound, we compose all of our outreach, but the human reviews and sends. That's the second part. It's called Rox Outreach.

After that, when you do reach out, it does all your meeting work. It does your pre-meeting research, in-meeting support, and follow-ups.

Again, the mindset was—let's call it Obama, or a president. Before you go into a meeting, you usually get a brief, a folder with everything lined up. What we want to do is ask: What if every knowledge worker had a team of analysts who were basically preparing them to be the best version of themselves?

Why is this just for reps, though? This would be super helpful for VCs.

We have tons of VCs on the platform.

Yeah.

And 100%, we're pretty maximalist in terms of where systems like these could be helpful. We think deeply about what's going to happen to the knowledge workforce across the board. Today, we have some of the best VCs—hopefully, you as well—using us.

Awesome.

We tune our agents based on the context. The email that it drafts isn't really trying to convince a founder to take a meeting. Those are things that we have to do, but it's completely feasible.

Don't worry. I send Twitter DMs, as you know.

Yeah.

In a world where you have this infinite supply of outbound emails because you create it, and you're able to do so much more, does outbound become less effective when supply becomes infinite?

Yes, and yes.

Yes, it becomes less effective. Our view is: How do we make the communication authentic? A very basic thing is that the email needs to come from your email. Second, the email that comes out should be personalized in your voice and tone, and then you should probably review it. That’s the only way to actually rise above the noise floor, because the noise floor is increasing exponentially. Everything else just will not work.

The second thing is, you brought up multithreading and talking through it. The key thing is that the human has to create a human connection, and one email’s not going to do it. It usually takes 13 to 15 touches: emails, LinkedIns, meeting them in person, getting on a plane, as Jason says. Again, our idea is: How does this swarm of agents help you produce the content? But you have to really put in the work to have those touch points. If people think that you can close enterprise deals by having one email go out, that’s not going to happen.

You mentioned earlier how everyone’s spending in a prior generation actually impacted the quality of sales reps, maybe. There is a real belief today that we are seeing this massive contraction in spending. Are we seeing CFOs spend less and consolidate more?

Mm-hmm.

Consolidate more, yes. The definition of spend, what I’m seeing, is changing, and it’s changing from programs to people. When you look at traditional SaaS—and obviously, I knew New Relic, and I’ve been in this space for a while—we’re thinking about CRM, we’re thinking about productivity tooling, and basically selling from the IT budget.

Now a lot of sophisticated buyers are looking at systems, hopefully like Rocks as well, and saying, “Hey, can I just make my humans a lot more productive so that I might not have to hire as many?” Just brass tacks, right? So their view of the ROI is changing from ROI on just software to the people spend.

What we focus on is that, for right now, we only convert if we hit our ROI metrics. We run a 30-day white-glove, like Palantir, boot camp where we work with the CFO, the CRO, and the C-suite on 3 metrics. The first one’s dollar-normalized outcomes, whether that’s pipeline generated or progressed. The second one’s adoption. The third one’s activity. The system needs to return all 3. If not, you can fire Rox.

Sorry, what is it? Number 1? Number 2?

Number 1 is a dollar-normalized outcome.

Yep.

It could be their choice: pipeline generated or pipeline progressed.

What do you prefer they choose most?

Pipeline progressed, because these agents are really good at making sure a human can manage a larger book of customers effectively if they have tens or hundreds of customers. The agents are doing most of the back-office work. They’re doing the research, outbounding, prepping for meetings, and managing your opportunities so that you can actually engage.

Number 2?

Adoption. Traditionally, adoption of a lot of these tools is abysmal, like the NPS scores and all. We index ourselves pretty aggressively. We’ve hit 75% weekly active usage and more than 50% daily actives. We want to be the daily-driver application.

You want to be daily active, right?

Daily active.

Yeah.

We have an iOS app. We have Slack. We really want to craft the application that people use to do their work. We index it that way. If people are not using it, you can remove us.

What are the reasons people have not reached daily active usage with you?

Three things. The first is relevance. It’s an agentic CRM with these agents trying to do a lot of the work for you. It collects all the data, finds relevant opportunities to reach out to, and then tries to compose all of your outreach.

Because we got pulled out of stealth so fast, Rocks is still in beta, honestly. It’s public, it’s open, but we’re still pretty early. Finding relevant moments to actually reach out to customers is a pretty hard engineering problem. Earnings coming out might be interesting, but it might not be compelling enough for you to actually reach out to those customers.

I find the most compelling moments are actually personal posts on LinkedIn. When it’s like, “Hey, my daughter just graduated from Brown, and this is a picture of the commencement,” and you send an email like, “I remember my day graduating at Brown.” They’re like, “Oh, wow, I like Ishan.”

Exactly. You kind of hit on it. You should be our head of product as well. Give us a lot of feedback. We’re moving from company-level insights to contact- or human-level insights, and that’s the journey. That’s the number 1 piece.

The second thing is that it is a net-new form factor, a net-new UI. If you look at our product and UI, it kind of looks like a CRM, but it’s actually a swarm of agents working for you. So there’s a new paradigm of UI or UX, which is between a chat interface and a traditional flat SaaS experience, where it’s more task-oriented. That’s where we’re at.

The history of Rocks is that it’s a pretty experienced team. The first 3 to 4 months, we built a guard mode agent that knows everything about every customer, with a chat interface. This just does not work.

Do you think a chat interface is the right interface for the next wave of AI and AI adoption?

I’m pretty pragmatic, because we have to be very iterative on the interface layer. We’re very opinionated about the back end and how to manage all the context, but on the front end, we just have to really track consumer expectations. What we learned was that chat interfaces are very powerful, but they’re very limiting for doing daily work.

Why are they limiting for doing daily work?

Our end users expect the software to do work, so they essentially want workflows, right? If the agent went out and did customer research, they just need all the customer research listed, versus asking, “Hey, tell me this, tell me that.” So we have account reports and org charts pre-created for them.

Similarly, for outbound sequences, instead of you going and saying, “Hey, produce an email,” we just queue up all of your email tasks. These are places where the software does have to lay out its outputs. Chat is an interface for them to do more introspection and go deeper.

I’ll give you a very simple example. First, we start with a chat app with just a blank Google bar. Next is, “Give me pills. Give me suggested things to click on.” Then you hit pills, like, “Hey, give me more output panels.” Those things work really well in consumer applications, like Perplexity, which is an amazing product.

But in B2B software, where people are coming in to get a job done, we have to do more. So what we did is decompose the experience into essentially workflows designed for the job at hand. We went with a jobs-to-be-done style: What are the jobs to be done? How do we organize it that way?

Meetings are separate from emails. Emails are separate from account research. The adoption was insane: We went from a few people using us to the whole company using it very, very fast. Again, is that a terminal state? I’m not sure. I think we’ll be iterative on the UI and UX.

You mentioned adoption there. Are we about to see a wave of the most unprecedented churn in enterprise AI tools?

Churn and retention are the biggest things that keep us up at night.

I think it’s actually your 6-month-old, dude, but sure.

Yeah, exactly. We’ve chosen the hard path: We want to be on the main workflow train and build applications where people come in and do important work every day, which is not easy.

I think where people are going to die off, in terms of usage, is if they just lose trust. What you have to understand is that these products are inherently nondeterministic. These are probabilistic systems, right? The experience and the research output that you get for, say, your Siemens agent could be different from what I get from the Siemens agent because the LLM is trying to figure out what’s relevant to me.

If the end user sees something that’s not good, it’s going to shave away some of their trust. If it gets to a point where they’re like, “I just can’t trust this thing. I’m not going to come back,” that is the biggest thing. Is there going to be retention and churn issues from an end-user perspective? 100%. We’re seeing so many tools come in and out.

Do you think we’re still in the experimental budget phase for enterprises buying AI tools?

There’s a massive amount of experimental budget. We’ve made the decision not to tap into it at all, because the focus is on retention, usage, adoption, and ROI. We only work with customers from their operational budget—program budgets, now kind of people budgets. That’s the way for us to control it and be truth-seeking.

But we are in the very early stages of the adoption curve. What I see is a small group of outlier operating CEOs, COOs, and CFOs who really understand that there is an existential decision to be made this year: to transform the business and keep their lead.

The rest are still figuring things out.

When we think about bluntly giving agents to every sales rep and allowing the swarm effect so they can be much more productive and efficient, will sales teams be dramatically smaller in the future?

Yes, dramatically smaller. Just like engineering, just like any other function in any business. It could be 10% to 20% of current org sizes, and it’s a harsh reality, but that is a probable productivity gain path we’re on.

It’s tough, it’s brutal, and I’ve worked on automation systems all my career. I built robots in warehouses before—the company that Amazon acquired. We’ve seen these things happen all the time.

And the 10% to 20% that remains is the human-to-human interaction that gets on a plane to Dallas, comes and shakes your hand, and says, “Ishan, I can’t wait to work with you.”

That’s why I’m here in London, right? I think Jason’s point is important: those folks who are there to build nontransactional, long-term relationships at a strategic level. It’s a high-ACV, deeper conversation. They’ll be able to do a lot more.

They’ll be able to do 2 to maybe 3 times more this year. We’re seeing some of the best organizations increase their book size. You would’ve supported 5 customers last year; this year, maybe 7; next year, 12.

If I’m a founder listening today, what can I do to dramatically increase the sales efficiency and performance of my team with AI today?

First, hire the right folks who are future-proof and understand the world. Are they using these tools, maybe illegally, in their personal lives? If they’re not using them and don’t see the alpha in their personal lives, they’re probably not going to get it.

Second, ramp much faster because, A, you can do it faster; B, consumer expectations are changing too fast, and then you have to do so. Once they’re fully ramped, push the envelope on how many customers they can support. Those would be 3 things, without obviously losing overall quality and attainment levels.

Will there be teams that are dramatically smaller and pulling in the same numbers? I think it has to be proven, but I have a strong intuition that’s where we’ll get to.

Do you not kill data providers? When you look at your ZoomInfos of the world—respectfully, I don’t want to single them out—or your Apollos—

They’re all amazing.

—or any of the others, you don’t need them if you’re supported by an agent swarm.

We at Rocks, and these amazing companies working in the space, are deeply thinking about how to source agents. But the agents have gotten phenomenally good at gathering information at scale from the public internet.

Yeah.

Now, systems like Glean are doing that in the intranet, in the internal systems.

Yeah.

The agents will get more and more capable of collecting information, especially with search APIs and the Xs of the world. Having said that, data providers provide a source of value because they have 10 to 15 years of back-tested data: human-gathered contact information and LinkedIn information that they’ve been verifying through a lot of software, but also through a lot of operational rigor that’s really hard to replicate.

Henry in ZoomInfo, for example, has built a phenomenal business. They probably have the highest-quality contact data. They have a huge machine that they’ve built up to do that.

So you partner with them and use that internally?

When we go into a customer, our agents will enrich contacts. If you have ZoomInfo data in Salesforce or whatever, they’ll search for it and verify the information. Internally, we’ve OEM’d a few data partners, and that'd be Waterfall, like Clari, just to bootstrap us.

These agents today are enriching the information that you already have versus actually gathering it. There’s a split: company data is free game. Company data is public. Agents are pretty much at par.

Contact data is where the divergence happens. If you look at the question you brought up, it’s LinkedIn information.

Mm.

LinkedIn is a walled garden, rightfully so. It’s an extremely valuable network that Microsoft and LinkedIn have. The people who have LinkedIn data as a data product that they sell into are just in a different world that we don’t see ourselves going into.

We chatted before about Agentforce, and they are a core competitor. When we think about who wins in this next wave—whether it’s incumbents with distribution and a lot of touch points, as they do have, or whether it’s the Rocks of the world with speed, innovation, and agility—how do you think about that trade-off, and who wins?

It’s a brilliant quote from Alex Rampell: will the incumbent acquire speed before the startup acquires distribution?

Yeah, he’s spot on. Benny is the GOAT. You respect him a lot. I’ve learned so much from him. I used to read, read about his stuff way back in SmartUp in India.

Our approach is just time to value. Going back to the New Relic learnings and being a product-first founder, incumbent systems have data and distribution. They have massive amounts of both. But from our perspective, it takes quarters and system integrators to actually implement things and prove ROI.

For us, it’s about how we leverage our shipping machine to land something that gets started today and gives you ROI in an extremely time-compressed manner. Then we own the right to expand from there.

What’s going to happen is that all the incumbents have obviously repositioned themselves as AI companies or agent companies, and they’re also all bundling. As SaaS exploded, there was massive unbundling across sectors, definitely in the CRM space. You have phenomenal businesses like Clari and Gong and, and Clean folks.

They’re all bundling, and they’re AI companies. We’re in this bundling cycle. With us, we want to be the default new option of an AI-native, bundled software provider.

Now, when we think about getting into the field, it’s really difficult to build out a sales team, and I think the hardest thing is moving from founder-led sales to sales-rep-led sales. You’ve done this before, but you’re doing it now with Rocks. What are your biggest lessons in the transition from founder-led sales to sales-rep-led sales that you’re remembering today as you do it again?

Do you have a repeatable sales motion or not? And are you being absolutely truth-seeking?

What is repeatable sales motion?

Repeatable sales motion is being able to go from a list of potential customers to a high-quality first meeting autonomously, without the founder helping you. The second question is, can you go from a successful first meeting to landing the customer without extreme founder assistance? If you can’t do both, you don’t have repeatable sales motion.

In this early stage, where I do have to grow really aggressively, what I’m looking at first is, based on the case studies and reference studies, can an enterprise AE come in, build their own list of target customers without needing a BDR team or marketing people, and then go from there to the first meeting autonomously?

Can I completely take a step back? After M1, we just have to be truth-seeking. We don’t have a massive brand, we don’t have a huge demand-gen engine, and it’s not perfect product-market fit. We’re early—very early—and founder-assisted. That’s what we’re looking for.

The working session that I do is going to walk you through the whole strategy: here’s what’s worked, here’s how we messed up, and here are our reference customers. How would you go about producing your list of people to reach out to, and how would you actually get to having a high-quality first meeting?

Do you go for a sales leader who brings in the reps? Do you go for more junior reps who you train with that playbook? What’s that first step?

There are different variations. I back myself to recruit, train, and build a machine with 5 reps reporting to me directly, and then bring on a sales leader who doubles the team and takes it from there.

In those reps, do you want reps who’ve sold that size deal before or sold in that segment before?

Yeah. They should have held a number at that ACV size for 2 to 5 years.

Uh-huh.

Not less than 2, ideally not more than 5. Within those, they need to show that they’ve been innovative with the product. Within deals, they either built tools on their own or were creative in progressing or closing the deal, because early-stage selling requires a lot of creativity.

The founders can come in, and we can use investors, hiring, and all that to paper over cracks, but they need to be creative. The third thing is that they need to want to make money, right?

I really care deeply about the whole thing. You’d be surprised: a lot of AI companies’ AEs are not compensated with commissions.

Really?

Yeah, a lot of them, especially during COVID. I really try to get deep into it: did you actually hold a number? What did your compensation look like? Where do you want to get to? Because it gets deep into the psyche of a person.

The best AEs are complete owners. They're owners of their own book, not very different from a CEO or a founder. That's kind of the third thing I try to get to, and that separates a lot of folks who are doing it because it was easy. They also have a real estate portfolio on the side and are doing a lot of things, so you're trying to hone in on those three things.

First, they have some experience holding a bag—a real enterprise bag. Second, they're creative. Third, did you hold a number? I go really deep into that.

How quickly do you know if they're good or not good?

My interviews are long. I don't believe in 30-minute interviews. Either I trust my honor-based network of folks telling me somebody's good or not, or I do 1.5-hour working sessions. In the working session, by, I would say, the half-hour mark, I can tell based on the quality of the questions they're asking.

I'm coming into one of your working sessions. How do we spend that 1.5 hours?

I try to overshare context. The Silicon Valley ethos is: don't hide anything. Just share what you're working on, because if somebody beats you at your own game, that's on you. So I just share everything: “Hey, here's the product strategy. Here's where we are. Here are customer case studies.”

That's probably about the first 30 minutes. When I'm doing it, I'm trying to see the questions that they're asking. They might ask about pricing or case studies—what worked and what didn't work. That could take 40 or 45 minutes, and I do that for every candidate because I'm trying to understand how they would go about it.

That's where we want them to abstract it and build a list from there. Their brain should be running. After that, we get into specific scenarios. “Okay, now that we walked through the scenario, let's say we're working a strategic deal, a call with a large company in Europe. If I gave it to you, how would you run this deal?”

Then I flip it and ask them questions: “What if this happened? What if that happened?” The first 30 to 45 minutes is me sharing context and them asking questions. The next 30 minutes is them walking through a deal. The last 15 minutes is, if we feel it, talking about next steps. If not, we wrap early.

Do you do hiring panels? Are you happy to make a decision after that 90 minutes?

Yeah. I try to do founder-led, early-stage sales with customers and talent. You need to be able to make deals in the room. Again, a big difference between going back to, obviously, Dave at MongoDB and Amit at Datadog. If you look at these dialed-in machines, a lot of the time deals don't happen because the account team has to go back and talk to deal strategy.

The best teams are enabled with tools that let them make things happen if there's some heat in the room. So if a candidate has gone through it, I'm spending an hour or 1.5 hours, and I try to progress things along.

Reps get paid a lot, actually, in a lot of cases. What's your take on, “Hey, I'm willing to pay over market for the best. We pay market and then comp”? How do you think about lessons on that?

My framework is one of explainability. The idea is that compensation and how much money people make in equity should be explainable. I'm still old school. It doesn't need to be public, but let's say it did leak and became public. Everybody's compensation should be explainable. There should be a rational reason why somebody's compensated a certain way.

So I have explainability guardrails that I don't want to breach, because if you're giving somebody an irrational package and it leaks, it's going to have trickle-down effects on the team. The company is a team. The team's the most important part.

What I try to do is, if somebody's truly an outlier, they've gone through the process, and they're really capitalist and thinking about incentives, I'm very aggressive with performance-based compensation or equity grants. One hundred percent, the outlier victories happen when you break rules.

I totally agree with you, actually, on the opacity of salaries. I think people always view everything with the expectation that everyone's rational, when actually everyone's entirely emotional and human.

Yeah.

Which means they're irrational.

Yeah.

They're also volatile, so at some points they'll be like, “That's completely unfair. I just signed a massive deal.” Then, when they've had a dry spell, it's like, “Oh, it's actually a fair representation.”

Something that I'm worried about is you mentioned target setting earlier, and you mentioned Jason Lemkin earlier. I grew up with the Jason Lemkin school of thought for founders, which is triple, triple, double, double. We're like, “You know, this is what you need to do to get great rounds done.”

The truth is, now you've got your Macaws, you've got your Lovables, you've got your MidJourneys. These companies are scaling to $40 million, $50 million, $100 million—I mean, in 12 months. I'm worried that we've lied to a generation of founders about what it takes to get great rounds done with revenue trajectory. How do you think and feel about that as an AI founder today?

Probably the most interesting thing for me to manage my psyche around. The way I think about it is, you're building something long-term, and what matters is enduring revenue—high-quality revenue from customers who can be with you for the long, long term. Are the unit economics and the numbers looking great for that?

For us, it's traditional enterprise land and expand. Are we landing the right logos? Are we landing them at a decent land ACV? Are we growing?

What is a decent land ACV? I obviously don't know. Is it $100K, $200K?

Yeah, $100K—low six figures—and then really prove value. We're doing a no-brainer, quick land. We compete on time to value. Is there a path to high six figures or seven figures? That's what we really care about.

What do you want to see in terms of that sales cycle and then expansion time?

If it's upper-mid-market enterprise, we want to be at 45 days from M1 to having that initial land done and signed. After that, it depends on how we construct it.

There's a 1-year contract with a 30-day paid evaluation period. We have these discussions with the customer, and we're still in beta, right? Enterprise customers are buying beta software, so we give them the right: “Hey, after 30 days, if you don't have the ROI, you can downgrade to our premium product, or we can have a conversation.”

After that, it converts. Once it converts, we're looking to deliver more value and drive more expansion. So it's 45 days to get to the first land, and then the clock starts. First, we need to prove value in 30 days. After that 30 days, it's about delivering more value and finding an expansion.

I think we've already at that six-figure range, and the question is how we're going to really grow from there. Ultimately, we want to be the essential software system that's helping you secure and grow your revenue. We want to be essential, just like SAP is or Salesforce was.

I totally get that in terms of the desire for enterprise sticky revenue and building enduring value in the long term, rather than getting caught up in sugar-high hype cycles.

I tweeted last night—it's really unfair of me, actually—that preemptive rounds are incredibly dangerous. They encourage you to do more stuff that you shouldn't, lose focus, be less capital-efficient, and then become unable to scale into a price that's too high. You get people leaving because they know that, and then the negative vortex continues.

You've taken preemptive rounds.

Absolutely. A few times.

Why am I wrong?

I don't think you're wrong. It's really, really hard, and I've gone back and forth. Ultimately, my framework and our framework is that we're building this for the long term. Who do we want around the table who can help us build this for the long term?

Did you put the cash in the bank for a rainy day?

We raised $50 million. I think we have—keep me honest—$47 million in the bank. We put in our own money at pre-seed, so we have full skin in the game. We're not going anywhere.

Bastard. Should've come to me.

Day in and day out, I've managed a billion-dollar revenue book. I've seen large businesses operate and been part of it, so it requires a lot of experience being able to manage capital well.

I think it requires the mental discipline. Honestly, what I see is young founders who raise the $50 million and then go, “Ah, well, we could do this as well, and then we could do this. Fuck it. Why don't we just do that as well?”

Suddenly, they do so much more, and they don't have the discipline to say, “Hey, I'm putting it away, and this will sustain us for 8 years if we want it to, so we can build with the long term in mind.” They just do too much and bring everything forward.

I make mistakes all the time.

What was the biggest mistake you've made with Rocks, between us?

Between us and everybody listening in, I think I spend 40% to 60% of my time on talent—the team. I underestimated how important and powerful it is. We built a founding team made up mostly of top AI engineering talent, and that team has built the product and the culture.

In hindsight, I would have moved faster to build out that team. Then, going back after I went to San Francisco, I thought, “I need to spend more time.” So the biggest mistake is that I would have maybe 5X’d the value I put on hiring.

The second is very detailed. When thinking about AI hiring specifically, in the first 3 months I wasn’t pragmatic enough. I was still looking for, “Hey, maybe we could do research and hire researchers.” No. We’re pragmatic. We’re building apps here. We’re building products, so let’s work with the best intelligence providers: OpenAI's of the world, Anthropic's of the world, and the Gather of the world, and let’s centralize our focus on working with them to build an amazing application.

By about month 4 or month 5, we learned that we should hire amazing applied AI engineers who are shipping products. We’re not a research company. I would say those are the 2 things.

I’d have game designers on your product team. If you’re thinking about the weekly and daily, most importantly, I would just think about, how do I create retentive loops and leveling? How do I create this addictive desire to continue to be on the platform?

Actually, you can see it especially in sales teams, which are highly competitive, where you can really create these kinds of leaderboard-style boards and real competition dynamics.

Absolutely. You’re spot-on. I think that is the biggest learning: getting designers early, people deeply thinking about experience, and getting product engineers who are really thinking about the application. Add some gamification post-product-market fit, and really double down on that. Get there faster.

We’re a year in. We’re doing okay, but I would have been more opinionated on day 0.

Listen, I want to move into a quick-fire round.

Let’s do it.

I’ll say a short statement. What sales tactic has not changed over the last 5 years?

Jason guessed it. Get on planes.

Yeah.

Meet people.

Totally agree with that. What sales tactic has died a death? No more.

Automated emails.

I totally agree. Tell me about a deal that you’ve closed where you’ve done something really creative, outside the box, to get it done.

I helped a buyer map out his path to promotion at a large streaming company, which had nothing to do with the technology that we’re providing. It was about deeply understanding his career objectives to help him be successful. Nothing political. He was just looking for help. I probably spent 2 months on it.

He got promoted and then didn’t do the deal.

He did the deal.

He did the deal?

And he’s running the whole show now. It’s a pretty large public company.

Oh, wow. That’s amazing.

Yeah.

You said earlier about your love for Perplexity. I love it too. Does it survive in a dominant monopoly world of ChatGPT?

If they keep the eyeballs and keep shipping the features they’re working on, I think they’ll continue to be a player. Is ChatGPT the default? Yes.

Do you not think it follows the same market dynamics? I’m an investor in Perplexity, by the way, so I’m a big champion. But—

Arvind talks about it as well. It’s really hard to build what they have built and the talent that they have. At the very minimum, they’ll find a massive niche, like large domains—

Shopping.

Yeah.

Finance.

Yeah. Large domain segments that they will be the best at. If not, they would be a strong player in the top 3 for sure.

What will be the most significant structural change to sales teams in the next 5 to 10 years?

A change in their quota. As humans get more productive, as they’re able to support more and more customers, the book size that they support fundamentally changes. If that changes, the complete structure of the organization changes.

Instead of having large teams, you need fewer teams who hold more book. Management layers are going to compress, and then you need different kinds of leaders. That completely changes. Sales and marketing are 40% of your OPEX, right? It could potentially go down to 20%.

Where does that shift to, then? If it goes from 40% to 15%, that’s not going to be a 25% difference in technology spend. So where else does it get redistributed to? Are they just businesses with much higher margins, much healthier businesses?

50% of the customers that we work with—this is boring stuff—are private-equity-owned. Yeah, 50% of the people who come inbound are portfolio-wide rollouts. So there is definitely an element of efficiency. Can businesses run in a dramatically more efficient manner when they’re not growing at all costs? Yes. There’s a huge part to it.

Where does the capital go? Hopefully, some of it goes to increasing the compensation of your supercharged frontline sellers. The other thing—I hope, and I’m kind of confident—I still do think there’s so much more to invest in R&D.

I think if you’re a technology business that wants to be a winner or be number 1, R&D is still severely underinvested in most scaled businesses. I hope your frontline quota-carrying reps gain time to make more money as the company’s doing well, and that capital gets reallocated to R&D, because we don’t do enough of it.

How do you maintain morale when goals are missed in a sales organization?

If you’re running the motion where everybody’s hitting quota and the company’s doing great, the morale hits happen when the company collectively isn’t doing great because of something macro or the TAM. If it’s the other situation, where there are winners and losers constantly, there are 2 different things.

In general, my view is 3 things. First is: keep cranking. You’ve got to keep working. Second is: truth-seek. Call your gaps and wins early. Take extreme ownership. Talk about what worked and what didn’t work.

The third, in terms of morale, is that I do believe in the power law. There are going to be those 10% to 15% of folks who are true believers and are going to be there for the long term. Just double down on them. If they hit, great. If they miss, how do we help them win the next time?

Then the middle 40% to 50%—will they sway to the top 10% or the bottom 10%? That’s what you have to manage, but the only way to manage that is by doubling down on your outliers.

What do you think Salesforce has done really well with Agentforce that you take and learn from?

Everything Marc Benioff has ever done. He’s a phenomenal visionary and has really told the story well. Marketing and sales are obviously top-notch from what they’ve done.

They have taken legacy enterprise buyers—think CIOs in European and Asian industrial companies—and really gotten them to go on this agentic journey, where every CIO in the world is buying this year, right? A lot of it is because of him and Agentforce.

So painting the vision of what an agentic workforce could look like, he’s done a great job. That’s why he’s the GOAT, right?

Which other company’s sales strategy have you been most impressed by recently?

I think, from a recency perspective, if we call it the last 5 years, I would say MongoDB, Dev, and that whole crew. I see their leaders in every great company. Ex-Mongo people running go-to-market in AI companies or fast-growing companies is so common.

They just build a winning culture, and he’s completely transformed that business. It’s a huge success. I learned a lot from him.

The other one would be Ramp. Eric and those folks were our second private beta customer. Now we have one of the most-used products. What I’ve seen there is really high-horsepower sales folks who’ve grown from selling 1 thing to selling multiple things and now going upmarket, selling to some of these largest customers.

It’s all because of this core talent base that they had. Those 2 things.

Dude, listen, I’ve so enjoyed this. I love shows like this where I can bluntly just have a very fluid conversation—

Yeah.

—and you do not stick to schedule. You’ve been fantastic, so thank you so much.

No, I appreciate you for having me in. From every founder and every operator, thank you for doing what you’re doing.