AI主导的销售：1Mind的超级人类如何驱动指数级增长——《Agents of Scale》播客

Hello, and welcome back to the Cognitive Revolution. Today, I'm excited to share a special cross-post from Agents of Scale, a new podcast from Zapier CEO Wade Foster, who long-time listeners will remember from his September 2024 appearance on the Cognitive Revolution, when we discussed Zapier's evolution from no-code pioneer to an early LLM knowledge worker. Wade's guest for today's episode is Amanda Cailo, previously founder and CEO of Unicorn Six Cents, and now founder and CEO of OneMind, a startup that recently announced a $40 million fundraise to support its mission to redefine go-to-market with what Amanda calls AI superhumans, video avatar digital teammates that engage, qualify, and convert prospects into customers. The company promises AI-led exponential growth, and as you'll hear from Amanda, the new era of AI-led sales begins at home. Adopting a classic dogfooding strategy, which Amanda refers to as drinking our own champagne, OneMind leverages its own superhuman named Mindy to source 76% of the company's pipeline opportunities. And when OneMind was working on a deal with HubSpot, instead of personally answering questions from the 20-plus people who needed to sign off on the deal, Amanda directed them to Mindy. Today, HubSpot is a reference customer, reporting a 25% increase in revenue after deploying superhumans in their product-led growth motion. And Amanda argues, with clarity that only comes from someone who's already lived it, that traditional organizational structures, with marketing, sales, and customer success functions operating as separate organizations, are quickly becoming obsolete. If an AI superhuman can support a buyer from their first website visit through an eventual upsell, businesses don't need to organize around human capacity constraints or other limitations, and can instead build a single cohesive team to support the entire life cycle. When it comes to the impact on jobs, Amanda is both clear-eyed and visionary. She sees account executives as safest from AI, because the challenges of building relationships and navigating organizational politics remain hard for AIs. But she warns that SDR roles focused on qualifying leads and solutions engineers who primarily do demos are likely to be outperformed by systems that can keep a prospect's full history in their context window, speak fluently in any language and about any industry vertical, and instantly retrieve any slide from thousands of decks. To ensure that OneMind is poised to take advantage of this opportunity, Amanda has a remarkable policy. If a human employee can manage to eliminate their own role with AI, she will reward that individual financially and find them a new role at the company. And if at some point that no longer makes sense, Amanda hopes that society will respond to AI-provided abundance by deemphasizing the expectation of economic contribution and elevating work that serves human well-being in other ways. This episode is brought to you ad-free thanks to sponsorship by Zapier, and I do sincerely encourage you to subscribe to the Agents of Scale podcast. Another recent episode that I particularly recommend is Wade's conversation with John Nerona of Gamma, which I'm going to take a little credit for identifying as the best AI presentation maker way back in August of 2023, and which recently announced a fundraise led by Andreessen Horowitz at a valuation north of $2 billion. If you're anything like me, you'll want to hear how they did it, and Wade and John's conversation about Gamma's approach to both product development and company building is every bit as valuable as this one. With that, I hope you enjoy this masterclass in first principles thinking about AI transformation with Amanda Cailo of OneMind from Wade Foster's new podcast, Agents of Scale.

It’s a whole new world, and an exciting one, so we have to reset the playbooks again. I thought, “Now I know what the SaaS framework and go-to-market are.”

I think we’re just walking into a whole new world that none of us know how to approach. Well, yeah, I was going to say: is there a playbook? Maybe it felt like there was in the last decade, but if there was, I never quite cracked it. At Zapier, we had to figure it all out on our own. Now, there’s definitely not a playbook for how to do this with AI.

How are you doing it?

Am I allowed to turn the question back to you?

Gosh, that’s a good question. We’re figuring it out.

Hey folks, welcome to Agents of Scale. I'm Wade Foster. I'm the co-founder and CEO here at Zapier, and my guest today is a serial entrepreneur, uh Amanda Cailo. Uh she is founder and CEO of OneMind. Before that, she founded Six Cents, grew it into a billion-dollar-plus company. And at OneMind, she is building superhumans, uh designed to sell, reason, and connect with customers. Uh kind of like your top salesperson, but but smarter. So, we're going to talk a little bit about what Amanda has learned building, scaling companies, but particularly now that she's doing it in the age of AI. Amanda, welcome to the show.

Thanks for having me, Wade. It’s awesome to be here. I always like to clarify that I didn’t take us to the $5 billion mark. I was there in the early years, stepped down, and brought in an awesome team. They’ve done wonders with the company, so I don’t like to take full credit for what wasn’t my whole doing. I did have a piece of it, but not the biggest piece.

I want to go back even further to kick things off. Tell me about CI Insights. This was the first company, I think, that you started and ran for a good while. What did you learn about starting and running a company there that you’ve now applied to your later companies?

That’s crazy. No one ever asks me about that company. I started that company when I was 22, so I guess I’m what you would call a serial entrepreneur. I was actually working for a company called Giga Information Group, which was bought by Forrester. I was working on a project, asked for a raise, and was told I needed to wait 6 months to get it. I thought, “I’m going to spin off and do my own thing.” That’s how it spun out.

That went on for 16 years, and we were basically in big data before big data was sexy and before it was really a thing. The younger generation probably has no idea what the term “big data” is anymore. We don’t say it anymore, but we were taking unstructured data, matching it up with structured data, and making sense of it.

Our customers included Cisco and Intel. We had a lot of big enterprise B2B companies, and I was merging their sales data with their marketing data and their customer success data and doing a lot of backward analytics. It was a good run, and it was a lifestyle business.

I never really sought out to be a venture-backed CEO of a tech company. In fact, I was living in San Francisco, and I don’t even think I knew what it meant to be venture-backed at the time. But I knew I wanted to build purpose-built software based on a project I’d worked on at Cisco. I thought, “Fuck, I need money to do this. How am I going to get $10 million to build what I need?” That’s how Six Cents was born out of that company, which had gone on for quite some time.

Got it. Now, with OneMind, you started a company post-ChatGPT and also started multiple companies pre-ChatGPT, so we get a little bit of a comparison here. What’s different for you this time around with this company? What are you doing differently, and how are you building your company now that you have these capabilities to tap into?

I think everything is different. Personally, I’m in a different place in my life, and one of my favorite sayings is, “The more you know, the more you know you don’t know.” Back in 2012 and 2013, when I started Six Cents, I thought I knew everything, and I knew basically nothing. I was going in blind.

Now, having been to this rodeo and seen it before, I’m coming into this world at Six OneMind from my past life at Six Cents. I made every mistake in the book at 6sense. Back then, there really weren’t any playbooks for go-to-market and SaaS. We were just getting started in the early days.

The only company to look up to in those early days was Salesforce, so we were all trying to emulate Salesforce, which was a mistake. As an early-stage company, trying to copy what they were doing without the resources they had was a huge mistake. Bringing people over from Salesforce and thinking, “If they’re doing it at Salesforce, they can do it here,” was another big mistake. I made plenty of mistakes along the way.

To your question about having the foundational models at our disposal today, I think it’s a world where, obviously, everyone is saying this, we can do so much more with less, and everything is about speed. At the end of the day, we have to ship fast. I expect a lot more, and I drive really hard.

I feel like I’m a combination of this empathetic, loving person who cares deeply about the people who work for me, but I also drive very hard and have incredibly high standards. I’m very transparent about where I stand on things. It’s a whole new world and an exciting one, so we have to reset the playbooks again. I thought, “Now I know what the SaaS framework and go-to-market are,” but I think we’re just walking into a whole new world that none of us know how to approach.

Well, yeah, I was going to say, is there a playbook? Maybe it kind of felt like there was in the last decade, but if there was, I never quite cracked it. At Zapier, we had to figure it all out on our own. And now, I'm definitely like, there's not a playbook for how to go do this with AI. So how are you doing it?

[laughter] Am I allowed to turn the question back to you? Gosh, that's a good question. We're figuring it out. We did the PLG thing for so long, and now we're trying to figure out how to turn that into an enterprise sales motion. I don't know that I've got enough lessons learned yet to pontificate on it. The more you know, the more you don't know definitely rings true to me on that question.

But you are building OneMind, which helps people sell better using what you call superhumans, which other people might call agents. What are you learning from your customers about what is working at this moment in time? We hear all the hype and fantastical stuff, but what is practically working? If you're running these companies today, what is the thing you should be doing that maybe you aren't yet?

Yeah, and I'm going to bring it full circle to what you were talking about with your PLG motion. First, to answer the question about what we're doing: we're building what we call go-to-market superhumans. It's a face, a voice, and a double-click on the brain. So think of it as a go-to-market Glean, if you will. It's all about being able to take action and have agency, and to have human-like experiences.

The face is a gimmick to get people to talk to it. I don't believe the face is where the core IP lives. I think it's really all in her ability to carry on a conversation exponentially better than a human. These superhumans behave and take on roles across your entire go-to-market cycle, everything from first touch and inbound. When somebody comes to the website, it's obviously not just a chatbot. She can share slides, give the pitch, give a live demo, read the screen, and see what's happening.

Humans have limitations. If I were giving you a demo right now and I were going through slides or a slide deck and you asked me a question, I might have the slide, but I can't bring it into the conversation in real time. I can't take 10,000 slides and be like, “Boom, there's the one that answers your question,” because I don't know where the conversation's going to go.

I think our superhumans supersede the capacity limitations of humans, both on memory—humans forget everything—recall, and the ability to go deep and wide. A company like Databricks, for example, that we're talking to, sells to everybody who needs data, right? Anyone who needs data across any industry, any vertical. Your salespeople can't really understand all the different industries and verticals they're selling to and talk the vernacular of their customers.

But what if you could truly empathize with your customers? We're building these human-like experiences that are better than humans, help you grow, help you cut costs, but at the end of the day, delight buyers. Create a better buying experience instead of going through a bunch of SDRs. No offense, but the 22-year-old kid who doesn't know your business is just trying to qualify you. Then you get passed to an AE who can't go deep technically. So then you get to a solutions engineer, a sales engineer, and then you buy, and then you go to a customer success person. Oh my God, it's a nightmare what we do to our buyers.

But to your question about where we see the most value and where we see that this is actually resonating and working, the answer is in those areas where there is no business model to put a human. For you, in a PLG motion, if you're trying to go sales-led, I would say, “Let us help you with what we call AILG, AI-led growth,” where your buyers actually want to have a conversation and know how to use your tool.

We're a happy customer of your product, and we use it within our product and absolutely love it. But I can tell you right now that I barely understand it in the sense that I'm not using it every day. If you're trying to sell to me, you'd have to dumb it down to the CEO level, right? Help me understand: “Okay, this is how you can connect your different systems, and these are all the different workflows they can do.”

If you were selling, you don't know if I'm going to be a $500-a-month deal or if I'm going to be a $5-million-a-year deal, right? By ICP and other attributes, you don't know those things, so you can't put humans on every one of those deals. But what if you could engage in a human-like way, have a conversation, understand my pain points and what I'm trying to do as a business, how I'm trying to scale, how you can use this to help us get to market faster, and how we can use your tool—and have that conversation for the $5 deal as well as the sales-led enterprise deal?

In those PLG-like sales cycles where we can't talk to everyone, where you have this massive long tail and you're trying to convert free trial to paid, put a superhuman in to engage your buyers, move them along their journey, get them using your product, and start paying for your product.

I think that's the place. For example, HubSpot is one of our key customers, and they increased their revenue by 25% when they put a superhuman in their PLG motion. Another very large—I'm not allowed to say the name—but a very, very large, let's call it, social networking platform is using it, and its sales cycles went down from 22 days on average to 2 days, and we doubled its ASP. Massive impact in that down-market commercial segment.

But we're also working for companies that are sales-led and doing enterprise as well, and helping support sales cycles.

One of the things you brought up that resonated with me is this: when you're talking to a human salesperson, they can't recall the hundreds of decks you have on a particular topic. They don't know all the industries. They don't know all that. Whereas for an AI, that's something it's particularly good at.

My co-founder Mike runs this thing called Art Prize, where he studies these things around AIs. One of the things he talks about a lot is that there are things AIs are very good at and humans are bad at, and then the inverse is also true. There are things humans are good at and these AIs are so bad at. It's kind of weird for us right now because we can sort of feel both of these things happening at the same time.

So I guess the question is: where does that apply for this product category right now? What are the things where you're like, to your point, the slide-deck recall, the things it's really, really good at? And then what are the bounds of it? Where are the places where you're like, “Hey, if this is the thing you're trying to solve, we're just not there yet”?

Yeah, great question. I think, to be honest, across the whole continuum—if I look at it, if I answer that question as it relates to the different roles across the organization—I think the AE is the safest at this moment, because there are a lot of those softer skills, right? Navigating the organization and understanding who you're talking to.

But I would say 80% of the deal is conveying information, understanding the buyer's needs: What is their pain point? What is the solution I can offer to solve that pain point in the most cost-effective, reliable, and accurate way? How can I get it into their product or into market to solve their pain as soon as possible? In that transfer of information, AI is exponentially better than a human.

People ask me if AIs will hallucinate, and one of my favorite things to say is, “Do your sellers hallucinate?” And they do, so nefariously. They know they're hallucinating to get the deal done, right? So they're likely to do it exponentially less. On the softer skills, of course humans are still here to meet buyers when they're active and ready to engage, and answer their questions. That's where I think the superhumans and agents can do exponentially better than a human can in that process.

You mentioned you think the AE role is the one that is most valuable still at this moment in time. I'm curious, when you look at these customers—HubSpot, the social network, even at 1mind—what about the go-to-market org structure is changing? How are they being built differently in this new age?

Yeah, I think right now, the playbooks we have are that we have a marketing team. If you look at the journey, the life cycle of a customer, you have marketing, sales, and customer success to onboard, upsell, cross-sell, et cetera. I think as we have these agents and superhumans—whatever it is—supporting buyers from step 1, from first touch all the way through to upsell, you're going to have to have 1 team managing that.

It's not going to be bifurcated into these 3 different departments that are in silos, that kind of talk to each other on an update call but aren't actually working together and don't have the common goal. I think it'll really come around to a very customer-centric, if you will, life-cycle management of the customer that's all in 1. The role of the human would be to manage from beginning to end, and I think we need new playbooks and new frameworks. I don't know exactly what those are.

I think we need to rely on smart people like Winning by Design, or whoever it is, to say, “What is the new playbook in today’s world, where all these tasks and the transfer of information can happen by agents and superhumans? And then where do humans play a role?”

The humans will play a role not in these siloed organizations, because we stay in silos because humans have capacity limitations. They have time limitations. They have recall limitations. They get sick, and when they’re good, they’re gone and they move to your competitor. All those things go away if the superhuman is doing the job.

Once you train it up and she’s good, she’s never going to quit on you—unless you don’t pay your OneMind bills.

Right. Well, how are you deploying your AEs differently from how another company would right now?

Yeah. Our superhuman—in fact, she is literally like, “I’m out for a raise right now.” Our superhuman and I have all these metrics really clear right now: 76% of the opportunities in our pipeline have been sourced by our superhuman. Our growth has been pretty fantastic over the last 18 months.

She’s not just the first touch, and she’s not just creating the opportunity, but she’s shortening our sales cycle. For example, when I was selling to HubSpot, I sold to Kip, who’s the CMO, and Kieran, who’s the head of AI at HubSpot. But their entire teams had to get on board with this. They had to say, “Okay, we’re going to implement this.”

The deal wasn’t done when they bought into the concept of buying OneMind. I had to go and talk to 20 other people. But instead of talking to me, I said, “Hey, talk to Mindy if you have questions.” I had marketing operations people asking data integration questions, so I was like, “Just go talk to our superhuman. Go talk to Mindy.”

I could see these conversations happening, and I think there were about 50 conversations with Mindy in that deal cycle. What probably would have been a typical 90-day average enterprise sales cycle went down to 30 days because they got their questions answered.

Then I went into that call knowing what they cared about. I was like, “Do you have any further questions? I saw you asked about our APIs. I saw you asked about how we integrate. I saw you asked about hallucination.” Then the deal cycle was like that, and it just completely shrunk that cycle.

We’re using it not just to get into deals, but I only have so many resources. I only want so many salespeople. I want salespeople to manage the relationship and get it over the line, and I want our superhumans to be selling—to be constantly, always on, always selling.

We’re building a superhuman right now to onboard our customers. That’s the next piece. She’s going to take them through, “Here’s what you need to do to onboard.”

Then we have a superhuman to support customers post-implementation. “Okay, you’ve got questions, or you want new data, or you want to keep your data fresh. How do you see into her brain?” Here’s a superhuman to support that.

I want to put as many superhumans as I can across the full life cycle and eat our own dog food, or drink our own champagne, or whatever you call it.

That is fascinating. I’m curious about the human psychology side, the buyer psychology side, of that. When you said, “Go talk to Mindy. Go talk to Mindy,” in this case, I think there’s probably a curiosity where people are like, “Well, I guess I am going to buy this thing, so maybe I should go give it a whirl.”

But if you’re not selling Mindy—say you were just selling some other random software—do you think that same curiosity would be there to go talk to Mindy? Walk me through that element of it, because I’m assuming you see this with all your cycles, where people are like, “I want to see how this goes.” There’s a curiosity that drives people through the cycle.

Yeah, it’s interesting. I think there are some industries and verticals where, if you’re selling into tech—if you’re selling tech to tech, you’re selling to technologists—they’re going to talk to it.

I think the world is moving in the direction that we’re all talking to ChatGPT on our phones, right? We’re probably further ahead of it in the Bay Area and in the tech world than others, but the world is still going there.

I love it when people ask me if they’ll talk to her, because it’s just talking. I’m not asking you to use a tool. I’m asking you to have a conversation. We all know how to have a conversation.

We’ve got a couple of examples, with HubSpot being one and Owner being another. Owner sells to restaurants, right? They created a clone of Adam, their CEO, to help them with their grader. Once they come through, they can talk about the scoring before they come in and tell them how well they’re doing, so they can sell their product.

If you look at that business as well, those restaurant owners know how to talk. All they have to do is hit “Talk,” and then they’re talking, right? So it’s actually not using tech. It’s not as technology-forward, even though you and I know there’s quite a bit to make that happen from a technology side. There’s a lot of tech behind the scenes to make a superhuman, but not to engage with a superhuman.

HubSpot saw that 88% of their audience, when they landed on the page where she lived, talked to her. Of those, 35% had a conversation that was deep and meaningful, meaning she was able to uncover pain points, solution with them, or drive them to a free trial or to purchase.

That’s wild. If you think about a chatbot, you’re lucky if 5% engage with a chatbot, and then maybe another 2% convert. It’s like nothing. All they’re doing is throwing you to content or booking a meeting.

I don’t understand why people love these inbound chatbots today. If you look at the conversion rates, they’re so low. They’re driving you to book a meeting with an SDR, which is like driving you to book with somebody who doesn’t know anything. Wouldn’t you rather just have a high-quality conversation right there, right then, than drive them to a 20-something who can barely represent your business?

Mhm. So, okay, you are replacing a lot of the traditional roles outside of the AE. Now, with this freed-up capacity, what’s next? Where are you spending your human capital, I guess, inside the company?

We have a thing at OneMind: if you replace your own job. I’m not there yet with the board, but I would love to get to the point—we haven’t fully papered this, just full transparency—where we forward-vest your equity or do something that allows you. If you fully replace your role, I’ll find another role for you.

If you’re smart and capable, then there are tons of other things to do, like building the AI agents within the company. There are so many things that we can do ourselves to make us more AI-forward within our own organization and outside of even just using superhumans, but also building agents to do workflows and tasks and everything else that we need to do.

I just think it’s a different job, and it takes a different skill set. It’s a different skill set. Some of those people may not have jobs.

I think we’re hiring for people who are capable of all of it, right? Capable of making that transition. That’s something we look for as we’re hiring right now.

But I think the world is going to shift, and I think we’re doing the world a disservice when we tell people that their jobs aren’t going to be replaced, because that’s not the reality that we’re moving toward.

That’s a real money-where-your-mouth-is move. I love it.

I think I feel similarly that the folks who figure out how to do this are creating immense value—immense value—for customers, for the organization, et cetera. But there still is that psychological safety, that fear. Humans feel loss aversion much more strongly than we do opportunity gain.

Even for someone who theoretically knows, “I can go do that,” there’s that thing that holds them back. Having these little nudges where it’s like, “No, this is a good thing. This is a good thing,” helps people keep developing their skills and pushing into areas where it’s like, “This is going to be way more valuable.”

Even though you’re not doing this task anymore, the next task will be just as valuable as the next one. You can repeatedly do that over time, and it’s such a powerful, innovative cycle.

Yeah, we have to teach ourselves to use new things, right? Anybody who doesn’t have that curiosity mindset—I was in remedial classes. I don’t tell this story often, but I’m happy to tell it.

When I was young, I actually thought I was behind. We moved a lot. My parents were separated, and I lived in 22 houses before I was 18. Because I was going into a different school every year, the schools thought I was slower than everyone else. And I believed it. I believed I was slower. I believed I was dumber than everyone else in the class. I couldn’t do it.

It wasn’t until later in life, in my 20s, when somebody took a bet on me, that I actually changed. That person passed away just recently, so it’s hitting me hard hearing this right now as I tell the story.

When I had that person take a bet on me and believe in me, I started to believe in myself.

And so I was like, “Wait, I can do a lot more than I ever thought I was capable of.” I think that’s true for everyone today. We can do, accomplish, and have anything we want in this world if we believe it with positive emotion and positive intention, and we truly believe it. But if you don’t believe it and you put it out into the universe, it’s actually going to backfire into the opposite.

So I think in this world, as we’re moving into this, we’re moving from carrying bricks on our back to having machines. This is the new revolution, and you either can embrace it and try to figure it out and be a part of it, or you will be left behind. I think it’s just really hard for me to hear people say, “Oh, well, no, it’s not going as fast as it is.” Those of us, like you and I, who really know what’s happening and are tapped in, I believe we do humanity a disservice by saying anything other than, “The job as you know it today is gone.”

Yeah. Thank you for sharing that. On this topic, you see a lot of your customers, prospects, and 1mind itself thinking more ambitiously about what is possible here. What do you think is separating those companies, teams, and individuals from the ones who aren’t? What is the 1% doing, and how do the rest of us cultivate that mindset?

Honestly, we’re typically selling top-down. In an organization, we’re selling to the CRO or the CMO, and they have a mandate from the board or the CEO to say, “You’ve got to use AI.” And I’m like, “Well, how do I use AI? I can put some copy-editing tools in. I can do some…” In go-to-market specifically, what are the use cases they can do with it?

I think they’re looking, and today is a unique time where you’re not going to get fired for trying something. I think 5 years ago, you might get fired for putting the wrong tool in place and it not working, right? So I think there is a mandate to try and do things and be ahead of the curve, because if you’re not ahead, you’re going to be left behind.

But if I’m being completely honest, there are some of our customers—even our customers—where people lower in the organization are blockers. They get in and see this as, “Oh my gosh, this might replace my job.” The fear factor makes them slow the project down, not get us the content we need, or try to put it in a corner. If you put the superhuman in a corner where nobody sees it, then it’s not going to get any exposure and it’s not going to work, right?

Those are real problems that we have to figure out and solve. Within that organization, I can’t change that, but I think it’s the executives at that company who have to make them feel comfortable that if they lean into this, they still have a job and it just might look different.

The executives you think are doing a good job of fostering that culture and enabling this kind of experimental mindset—what are they doing differently?

First of all, I don’t think it’s a matter of—I've seen a couple of companies. I actually heard about one company recently that had everyone go out and build agents, and they built thousands and thousands of agents, and half of them were all doing the same thing. And I’m like, “Oh my God, that sounds really scary.”

What’s cool about that, though? Actually, I can speak out of both sides of my mouth on it. I love the experimentation and fostering curiosity of individuals, but they’re not going to be production-ready. So I think there’s something beautiful about saying, “Go figure it out. Go find a task that you’re doing today and create an agent to do the task that you’re doing.” I think that’s amazing.

But are they ready to go into production and scale the organization? Probably not. It’s probably not the individual contributors who are necessarily going to do it. Sometimes they will, and that’s amazing, but I actually think the onus is on the executives to say, “Okay, what are we doing? Are we build versus buy? Are we doing it ourselves?”

I heard Megan Eisenberg, who’s the CMO of Samsara, recently say that she went out and built her own AEO within their company. They built the equivalent of Profound themselves, and it’s amazing and it’s working incredibly well. So I think there are places where you can build it yourself and do it. She’s encouraging the people below her in the organization, and there were many agents, if you will, that they’ve built, and then they bought some as well.

So across the whole organization, they had to reset and say, “Okay, we’re going to figure this out. I’m going to make this a mandate for my team to figure this out together.” But I don’t think it’s any one individual’s task, and I think that’s sometimes where people fall down and say, “Okay, go use AI in your day-to-day,” and have them figure it out.

Yeah, we all should know how to build an agent. We all should know how to build workflows and do things. But I think we have to, as leaders, guide and say, “What’s the best thing for us? Do we build it ourselves, or do we buy it?” In our places, I think there are advantages to both.

I think what we have seen on that experimentation front is that the organizations that do the, “Hey, we’re going to get everybody to build their first AI workflow or their first AI agent or something like that,” raise the floor of what’s possible inside the company. They may not necessarily raise the ceiling, but they raise the floor meaningfully, because now everyone in the company has this tactile feel for, “Oh, how do these things work? What’s possible? What’s not possible?”

And when you have that, the fear sort of fades into the background. It’s not that it disappears; it’s just that you have a much more practical sense of what’s going on here. You can see how you could go solve some problems, but you can also clearly see, “Oh, I’m still needed to help with this, this, and the other.”

And all of a sudden, you’re just problem-solving versus fear-mongering at that point in time, where the narrative with AI is that there’s a lot of fear-mongering going on.

100%, and I’d rather embrace it and be a part of change than be left on the sidelines.

Totally. So we talked about all these jobs changing. What are the new jobs that are arising? What are the new skill sets that you’re seeing folks lean into?

One of my favorite things—I don’t know what it’s called—is having an internal job to manage all the AI across your organization. I’m hiring for that right now, so if anybody sees that, we’re trying to figure out who that person is. Our CTO, our head of product, and our head of customer success have weekly meetings talking about all the agents and things that we need to build, and I’m like, “You guys need to do your jobs, and I need to find somebody to do this job across the whole organization.” So I think that’s super exciting.

Obviously, the prompt-engineering job—literally anyone can learn how to prompt-engineer. I find it interesting that everybody has these engineering titles, though. I was thinking about this in the shower this morning. I was like, “The forward-deployed engineer…” I’m like, “They’re not really engineers, and it’s actually not fair to the people who are engineers, who have worked really hard to get that degree and understand technology the way that they have.”

So I think it’s kind of cute, but it’s not what it is. Let’s call it something else that’s incredibly valuable, but it’s not an engineer. That’s not what they’re doing. I don’t buy into that. I put my engineers on a pedestal, and even in today’s world, where you can do things very fast, I think they’re one of the most valued—everyone’s valued—but they’re one of the most valued orgs within the company.

Of course, still in a world where you can do it fast. I like the word “builder.” We’re all building something. That to me feels much more representative of the work that is happening, and it is distinct from a true machine-learning engineer or something like that. That’s a different skill set in a lot of ways.

Yeah, and they’re creative too. What they’re not giving themselves credit for is that engineers are creative as well, but when you’re actually building these, there’s an art. It’s a combination of art and science. There’s an art to a lot of this, and they’re not giving themselves credit for the art that they’re doing, which obviously I love.

I have my own art that I do, and I think there’s a value in that. They’re devaluing the creative side of us, and I think in a world where agents and superhumans can do all the tasks and all the workflows, that creative-art, softer-skills side is going to be the most valued.

So you talked about this job of managing the agents. What does that practically look like for you all right now, and where do you think it’s heading in the next 6–12 months or something like that?

Right now, I think it’s somebody who understands the full organization. The agents have to talk to each other across functions, so it’s not just doing one task. Agents managing agents is the ultimate goal—the handoffs, the flows, and the process.

We're still trying to figure it out. I don't have all the answers, and I think the move—again, moving away from these siloed disciplines—is where the world is going. I think it's exciting. If I were starting my career again, that's where I would want to start. I want to understand the whole organization. I'm hiring a chief of staff right now who's going to have their hands in everything across each piece of the business, and I think it's similar. It's like a chief of staff for agents, if you will, for AI.

Yeah, the thing we're seeing is that the workflow looks a lot like a person who is reviewing the output of an agent, and that's kind of the job. Day in and day out, you're looking at it. You're saying, “Yep, that looks good,” or, “No, not quite.” If it's “No, not quite,” it's like, “Well, what if we tried this, or what if we tried that?”—to try to get it to more consistently output “thumbs up” versus “Ah, not quite.” So you're kind of managing these rubrics, managing the evaluations, and it really is like getting very good at just sort of holding a quality bar, knowing what that standard and rubric looks like, and then tuning the engine over and over again.

Yeah, the AI slop that's out there is incredible. I looked at something the other day and, at first glance, I was like, “Wow, this is fucking good.” Then I printed it out and read it, and I was like, “Ugh.” Do you subscribe to the dead internet theory?

I don't even know what that is. No, what is the dead internet theory?

The theory is that if you go to any of these major social networks, the vast majority of content out there is written by an AI at this point in time. What's being written and what's being read is all AIs talking to AIs. It's not real.

I mean, that's the world we're going to. Right now we're building superhumans to talk to humans, but it's agent to agent. That's the future, and so we have to be building infrastructure and figuring out what those processes are when agents are talking to agents. I can't even get my head around that sometimes. Like, fuck, is this going to work?

Well, I think we're definitely going to create a market for AI-free content or AI-free art. We're going to want that from humans. There will be a market for that.

All right. Well, then that's where I'll go sell my artwork.

I think so. I mean, it's the same thing as AI is clearly way better than humans at chess, but chess has never been more popular than ever. You kind of see that across industries where humans—we just, I don't know, we like to see what is possible by us. That doesn't mean that we still won't have AI and agents doing enormous amounts of economically productive work, but there are still a lot of places where humans—we just like each other.

And what if we move to a world where we all don't have to work so hard to meet our basic needs? Okay, I'm not going to get into my politics, but imagine Maslow's hierarchy of needs with basic needs truly met across the world, where we can all do the things that actually feel good to us, give us joy, and bring our energy up to another level. That's the world I get really excited about, right? If you and I want to create companies and do things that get us excited, we get to go do those things. But not everybody. Most people are in jobs that I would imagine they're not loving.

I was actually pushing back on a friend the other day. She's like, “I love it,” and I was like, “Really? What the fuck do you love about that? That sounds awful and boring.” She said, “No, I love this. I love what I'm doing.” I'm like, “I don't know. Do you love it? What do you love?” I love problem-solving and challenge. I love humans. I love people. I love managing people. I love understanding and connecting. I'm a relationship person and a product person, but that's what I love—not necessarily even building superhumans, right? I love that I get to build something that makes change. But what if everybody didn't have to work so hard?

Well, I think that's the direction we're on. If you looked at our agrarian ancestors, I'm sure most of them probably weren't in love with farming. If they looked at today's jobs by comparison, they'd go, “You all look like you're just having fun all day. You're a Twitch streamer, a YouTuber—is that a real job? That doesn't look real.” But here it is. We've created economically useful jobs out of what would appear to our agrarian ancestors as just fun or nonsense at the end of the day. So I do think that humans will just continue to move up that abstraction layer and find new ways to take up our time in ways that are economically useful.

Yeah, and I'll tell you a story. I have 2 brothers. Both of them were CEOs and entrepreneurs as well. I don't know—one of us should have ended up in jail if you look at our history. We had an interesting childhood. My oldest brother was a CEO, and then he just dropped out of this world of capitalism and decided, “I'm not in it anymore. I'm going to go be a Buddhist monk.”

He went and was a monk for a second, came back, and now he's a chaplain. What he does in his chaplaincy is walk around—or sit by people's bedsides when they're dying, when they have nobody else. He walks around homeless camps and says, “You need to talk to somebody.” There is absolutely no money in that. He makes absolutely nothing, but he's doing God's work. So I think it's a shift: we will shift the value of what our work is, and I think there'll be a value shift, hopefully. That's my hope in this new world: that his work will actually be valued.

Yeah, clearly that is valuable work that he is doing. You were talking about yourself: “I'm a relationship person. I love to sort of do these things.” Here you are changing the way that sales works. What would you tell a 22-year-old who's like, “I want to be in sales. I'm the same as you. I love relationships. I love helping people solve problems”? Where would you direct them?

I would direct them to focus on your softer skills. It's less about the process and methodology, the transfer of information, selling a product, and talking about a product, because I think all that transfer of information is going to go to AI and go to agents—superhumans, what we call them. It's more about creating connection to other humans.

One of the valuable lessons I remember my dad told me when I went to college was, “I don't give a fuck if you learn anything. Just learn how to connect to people. Learn how to have a conversation.” A really good friend of mine is from Ethiopia, and her family was royalty in Ethiopia. She told me that when she was 5, she had to go to parties and learn how to socially say hi to people and ask really intriguing and curious questions that invoked curiosity. I thought that was fascinating—that at that age she was taught, “This is how you show up, and this is how you engage people.”

So that's what I would teach them: learn how to engage people and lean into what makes you human.

On the “what makes you human” part, it's pretty clear to me that resilience is one of those things that is going to be increasingly in high demand. You have had an enormous amount of resilience, both in your professional and personal life. What do you think instilled that in you, and how do we cultivate that in others?

Ooh, as a mom of 2 young girls right now—I have a 3-year-old and a 5-year-old—I think about this all the time. I believe I probably have had the modest success that I've had because of the trauma that I've had as a child. I actually believe it's a trauma response. I love it when people tell me no. I love to be rejected. I'm like, “Hmm, I'm just going to use that as fuel to my fire.”

I've learned that if I feel something, I get it out, and then I'm over it. That is something I've had to learn in order to survive; it was a survival mechanism. But fuck, I don't want my kids to have to go through any of the trauma. They have a very sheltered life. My oldest daughter is adopted, and I look at the life that she would have had, like her biological siblings—a very different life. A 1-bedroom apartment in Philadelphia with her mom, who works at McDonald's. I would never want her to have to go through what I went through as a child.

So I don't know. The answer is, I don't know what the answer is, other than I try not to give my kids what they want. And I try to be a little bit of a tiger mommy. I don't actually believe in gentle parenting—not that you're asking me about my parenting styles here. But I also don't believe in anything abusive, either. I also don't know.

I’m struggling with it every day. So, if anyone has an answer, let me know.

I have three- and five-year-old girls, too. So, I’m figuring it out as we go, too. I think you’re onto something: of course, they’re going to have maybe more than I did when I was growing up, but you can build resilience into them by teaching them they have to earn things—not having instant gratification and all this other stuff—without, to your point, putting them through trauma at the same time. So, I have to believe that there’s a healthier path to still instill those same skills.

I hope so. My one thing is, for all my childcare—because I of course need a lot with the job I have—I always say to them, “The number one thing that will get you fired is if I see my kid crying and you give them what they want.” Do not fucking give them what they want when they’re crying. I don’t care how old they are: get it together, calm down, and ask me in a reasonable voice. But you’re not going to throw a fit and get what you want. That’s an instant no for me.

Yeah. It’s like, “Hey, we’re going to teach them skills here. These are teachable moments.” Exactly. You talked about how you like hearing “no.” You use “no” as an edge. Where has that served you well? Where have you thought, “That actually helped and fueled this engine of curiosity, this can-do attitude”? And where do you feel like it may have held you back, where you’re like, “I could have learned this in an easier way”?

Well, where that comes from for me is that, as I’m building a company and building a category, if you will, in the early days of 6sense, I got “no” all the time when I was out fundraising. Today, as I share it with the market and share what we’re doing, we get “no” quite a bit. We get “yes” a lot, but I actually get worried if we get “yes” too much. The reason is I feel like we’re too close to the reality of where we are today. I don’t think you create a category or create something new if you’re just working within the bounds of what people are used to today.

If I think about chatbots or something like that, somebody looks at us as a chatbot that’s slightly better, and I’m like, “Eek.” That’s not good. I don’t want to be slightly better; I want to create something net-new, changing how we’re doing business, how we’re going to market, and how we’re talking to our customers. I think category-defining companies are able to do that because you have to get “no”: most people won’t accept it because it’s not within the confines of the playbooks that they know. It’s a point solution, making something I’m doing today incrementally better, and I don’t want that.

Yeah, super interesting. This idea of hearing enough “no” to know that you’re taking enough chances. Correct. Do you operationalize that somehow, or is that just a feeling you have where you’re like, “You know what? I feel like we’re kind of taking it a little too easy right now, and we need to push ourselves to the next customer persona or whatever”?

Yeah, I don’t know how to operationalize it other than I love hearing it. Our salespeople, our reps, would call me and be like, “Oh, that was bad.” Or remember the early days? I brought on a head of sales when we were just at, like, 1 million dollars in revenue, or a little bit before. I remember when people would reject what we were doing, we were virtually kicking each other under the table, and she’d be like, “Ah, we didn’t get that one.” I’m like, “No, Katie, it’s good. It’s good. We cannot—if everybody loves this, we’re going to fucking fail. So, we have to get these ‘nos.’ These ‘nos’ are important.”

They tell us something, too. You should be like, “Why did they say no?” So, it helps us be better in how we can grow and create the best product that actually meets the needs of the market and where the market is. But I don’t think you create anything big by getting “yes” from everyone. That just means you’re like, “Oh, you’re slightly better.” And I get it. I don’t want everybody to get it.

Awesome, Amanda. Well, I think that is a fantastic place to end it. So, for those of you out here and hearing no, use Amanda's wisdom. It's just fuel to go figure out how to make it better. And thanks, everyone, for listening to Agents of Scale this week. Amanda, it's been a pleasure.

It's been an absolute pleasure. I'll see you soon. Thank you.

If you're finding value in the show, we'd appreciate it if you take a moment to share with friends, post online, write a review on Apple Podcasts or Spotify, or just leave us a comment on YouTube. Of course, we always welcome your feedback, guest and topic suggestions, and sponsorship inquiries, either via our website cognitive-revolution.ai or by DMing me on your favorite social network. The Cognitive Revolution is part of the Turpentine Network, a network of podcasts where experts talk technology, business, economics, geopolitics, culture, and more, which is now a part of a16z. We're produced by AI podcasting. If you're looking for podcast production help for everything from the moment you stop recording to the moment your audience starts listening, check them out and see my endorsement at aipodcast.ing. And finally, I encourage you to take a moment to check out our new and improved show notes, which were created automatically by Notion's AI meeting notes. AI meeting notes captures every detail and breaks down complex concepts, so no idea gets lost. And because AI meeting notes lives right in Notion, everything you capture, whether that's meetings, podcasts, interviews, or conversations, lives exactly where you plan, build, and get things done. No switching, no slow down. Check out Notion's AI meeting notes if you want perfect notes that write themselves. And head to the link in our show notes to try Notion's AI meeting notes free for 30 days.