Hiring when AI is a co-founder kind of sounds like a panel title that could be an SNL skit, but this is a legitimate thing that founders are thinking about as AI becomes more central to starting a company. So I want to know from each of you, and we'll start with Michelle: what are you hearing from founders as they navigate this moment?
Yeah, so I think there's a few different things, right? So, one of the hype things that we were hearing, I think, was that AI could solve all our problems, that we could replace all of our humans with AI. And I think the operating reality that we're living in today is just it's not exactly that. While AI can do a lot and it's really good at those repetitive tasks, what we're still left over with is someone to own really the judgment calls, the more strategic decision-making or thought partnering. And that's where you still really require a human. And so I think today what I'm excited to talk more about is just how founders are really getting this right and thinking through where to use AI versus where they still need a human in the loop, because there's definitely like a role design or redesign that we're going through today. And I think the founders who are getting it right are the ones who take that extra step and think: now that we've applied AI to all these repetitive tasks, these mundane kind of jobs to be done around this human, what does that new role look like? And then therefore, what is the profile that we need to fill that new role to actually be effective in this new world that we're operating in?
That's great. I was so we do a company-wide demo day every three weeks, and I was sitting there last week watching everyone demo what they had built, and that used to be an engineering-only presentation. Now it's anyone really demoing things they built, and I had this overwhelming feeling that was I have no control anymore. And I think when AI is a co-founder, or when AI when everyone is using AI, the speed is just it's like nauseating, and it's also exciting, and it's great, and it's terrifying, and all the things. And so I think the judgment layer that you mentioned, and the prioritization, and the alignment around where we're going just gets so much more important. We talk a lot about movement without progress and avoiding that at all costs. And I just feel like the amount of activity is so high. And so the direction you're going just becomes super important, and the clarity around the entire team and every node in your entire company knowing where they're headed and how that connects to the broader whole just becomes so important. So I think that's that's what it means for me.
For me, I would say first off, this is the most incredible time of my life to be a builder. And I know this is a room full of builders. So I like pinch myself sometimes with what's now possible to be built. At Gusto, we serve small business, and I like to think of the mental model as a teammate. So a co-founder is a type of teammate. And a great teammate and co-founder is something that's proactive, right? It's not just waiting for you to give it instructions. It tends to be more generalist. It's not just a single task or skill, and it's obviously accessible and available in many, many places. And so, you know, my example is going to be always small business. Like we have a customer in LA that's near the Dodger Stadium. It's a bar restaurant, and like they used to have to figure out how many people to staff for different games, and it was always based on their kind of best guess. In this case, leveraging AI, you know, they can do shift scheduling way easier, way smarter, looking at which teams are popular, which ones are getting more traction, who's going to have a bigger audience. So it mostly just means the small business owner is not alone anymore. They now have a teammate or a co-founder that can help them never be alone again.
I do want to go back real quick. What was it that made you feel like you were losing control, John?
Yeah, it's a great question. I think it is the activity. It's the velocity. And I think when everyone the floor has just been raised by everyone at our company, and we know that that's been raised at a lot of our competition or just generally in the market. And I think that as someone that I think probably does like to control as much as I can because I want us to win. It comes from a good place. Obviously not to the point where I don't want to suffocate our builders or suffocate the people that work at Leland, but I do want us to be headed in the right direction. I think looking around and just seeing, okay, I no longer can, and part of that is just company growth, but a lot of it, I think, is just the amount that each employee is now managing. If they're managing teams of agents, and those agents have agents they're talking to, I need the judgment for each individual node in that to be exceptionally high. So the caliber of talent has to go up too, because you're just entrusting them with so much more.
Have any of you managed a team of agents? What is it like? How do you do it?
I mean, I would just say to oversimplify, there's some people here probably go way deeper, like technology is a tool that gives you leverage. We call it agents today. Software should have always been giving people leverage. It always bothered me that a lot of software I used in my life felt like I had to adapt to it, versus if any of you have dedicated your life to building software, which is what I've done most of my life, like software was always there to try to give you superpowers. So specifically, you know, it means work gets done when you're sleeping. It means you can basically do 10, 20x more stuff more quickly. But what do you do, and why, and how do you measure quality matters just as much. So I don't know, some companies to me it's like, hey, if you can do, you know, the same thing with one-tenth the people, I jump more. Let's go do 10 times as much stuff with the same number of people. That's where I jump to.
Yeah. So you're like delegating the same amount of work, but with or more more work, but it's spread among the AI agents and the people.
We just think of it as we call ourselves Gusties. What's the impact per Gustie? And like the impact per Gustie is going up, up, up. That also means we can increase compensation per Gustie, because as you have more impact, obviously it makes sense to also pay people more.
Yeah. And on that, the agents they're they're only as good as how you onboard them, right? It's like similar to a human. Like if you hire a new person, they start next week, like you still have to take them through the onboarding. You have to explain to them what their job is and make sure you're being really clear in your instruction. Otherwise, they may not do their job how you intended them to do it, right? And so, to John's point, they they definitely give you superpowers, assuming you take the time to set them up thoughtfully.
Yeah. It feels kind of like, to use the superpower metaphor, like with great power comes great responsibility, because then even in my job, I'm like, well, I should be using AI agents so that I can write about them, because my job is to write about this stuff. But then I'm like, I don't know, like what if it like does something weird with my credit card, or like you see all the things on X where people are like, oh, it put my bank account info in Slack. But do you think those are like legitimate fears in a business setting, or like are those issues that are more with if you're trying to have it like book something for you? I broke my pen. It's fine. Don't worry about it.
Yeah, I think it's interesting, you know, as we think about agents becoming more and more like employees. And employees are, you know, basically what we do every day for work is we execute different tasks and we have a judgment layer on top that sort of determines how to do that task and which tasks to do. And the best employees that many of you have hired are very proactive. They try to figure out what they should do, even things that you don't necessarily know you want done. And any of you that have hired an intern that's super proactive knows that they like shoot and miss a lot, right? But you'd almost rather have that than someone that just sits there and looks at you and says, "What do you want me to do next?" And I think you're seeing a lot of the AI today go from like that intern that just sits there and says like just bully me and tell me what to do to like I'm actually going to go start doing things for you. And that comes with the security and the concerns around like, you know, I as an example, I so our company helps other companies become AI-native, and so we're constantly trying to do that to ourselves so that we can kind of transfer that knowledge to other companies. And so we're playing with all these different AI tools all the time, and I've really enjoyed playing with Grokbot, which, you know, now OpenAI has dots, and there's, you know, there's they're all going to, and there's obviously Muse and Instinct. And with Grokbot, it's really interesting because I have a I have like a primary agent that I engage with, and then there's like teams underneath of it, and there's teams underneath of that, and one of those like accidentally emailed like a few of my customers something that they it shouldn't have, and so now I'm like disciplining my primary agent, who's disciplining the manager, who's disciplining the employee. I'm like, what am I even doing? I'm like, I have an action company, and now I have this team of agents that that I'm running. But also like I've never gotten more value out of a system, an agent system, as I have out of this Grokbot system because of the proactiveness. And so it just comes at a cost. And I think that's where the judgment layer and the context that so many people, I'm sure, in this room and and others are investing in, so that the precision just increases.
One thing I'd add, just double down on the point about trust, which hopefully everyone can relate to. That's not an only AI dynamic. Like one of our core products is payroll. I would go back to like 2012. If you were to onboard to Gusto, then you'd be giving us SSN, bank numbers, routing numbers, personal data, company data, and then five days later, we'd be debiting like $10,000 from your bank account through a web browser in 2012. And trust mattered then. That probably wouldn't have even worked 10 years prior. Now it's just table stakes. And obviously, same paradigm in an agentic world, like who do you trust and why? Have they earned your trust? Have they abused your trust? And that's where, you know, that's a personal choice. There's not a single framework there. But you got to also figure out like what's the business model of this company. Are you the product or not? And then are they going to be around for a long time? Because if a company is facing challenges, they might start, you know, doing different things with your data to survive that you might not be comfortable with. So that's something I think a lot about is who am I giving data to, who am I trusting, but also are they a stable business or not? And if not, am I willing to take that risk?
Do you have any suggestions for founders on anything you've come across that makes using agents more secure, or things that you would look out for before using a certain product?
I mean, just a good tactic is sandbox and create demo accounts. Like even if you're going to connect email calendars, go create a fake email, a fake calendar. Like create a bunch of separate accounts, connect that, play with it, test, see if it makes sense, works. Kind of only have a one-time decision to like give an agent access to your entire historical inbox, which can totally make sense. I do in a number of situations. But folks that are hesitant, you don't have to just do it as binary. You know, go go sample, go go create fake accounts and have that be your test bed before you then jump to your actual core data.
And I think it's not even just, you know, how can you be sure you're choosing the right tools, but it's also on the internal side, like how are you ensuring that your team is using the information that it's providing and having good judgment on what it's actually trusting as accurate and then therefore sharing with your prospect and your customers, right? Something that, you know, I think that we've all probably been used to testing for at like the executive level when we're looking to hire new new candidates is some some form of like an ambiguous interview to test on like their judgment, right? So like it looks like you've set it up like a two-hour interview with this person. You're whiteboarding on a specific challenge. You want to see how they think through that challenge, and what you're looking for is like, yeah, hopefully they get to like a good solid answer, but you're you're testing for like do they have good judgment, good decision-making. And I actually think in today's world having something similar to that at every level it becomes incredibly important because otherwise this is your brand, right? And so if if AI told one of your employees, you know, this is data about your customer, but it wasn't accurate, and they took it and sent it out to them, like that could tarnish your reputation. And so I think it's not only, you know, are you choosing the right tools, but how are you enabling your internal team as well to make sure that they're they're making the most out of what they're they've been given.
So as you start incorporating more, let's call them AI employees or any sort of like AI infiltrating the company, do you notice it impacting the morale, and what do you do to manage that? The morale among the humans, not the I don't know if AI morale is a, you know,
So I can take that one to start. So I'm an advisor at Insight Partners, so I get the benefit of being able to look across a lot of different companies and see some pattern recognition. Prior to this, I was an operator as well at a number of early-stage companies. But one thing that we've been seeing across our companies is there's definitely an anxiety hit. Like on one end, yes, AI can do a lot of great things, and we'll continue to talk about that, but I think the reality is there's definitely an anxiety hit with a lot of the companies. And I think if you would have asked me this question last year, the question was typically or the concern was is AI going to take my job? And that was the thing we kept hearing over and over again. And I think today what we're starting to hear more of is can I actually keep up with AI and where my team is going and how quickly they're moving? Because as you said, that bar from that gap between those who are really leaning into AI and seeing these exponential results versus the folks who may not be leaning into AI as much, that gap is only getting wider. And so there's a lot of anxiety, I think, from all the way from your IC to your executive team on whether they're going to be able to keep up. And I think what's interesting is then like as this translates down into like an actual business, you're you're almost, you know, managing two different tranches of people, right? You're you're trying to manage those who are accelerating with AI while also managing those who are maybe falling behind but just not doing as much. And so it's, you know, it takes like an intentional investment to try to be transparent about what this new world means for your organization, for their role specifically, and then investing in enablement to try to bring the entire group along, even with them being at different places. But I'd love to hear you two are both in companies today, like how does that resonate? Are you seeing similar?
Yeah. I mean, I guess it is the full spectrum. I mean, what came to mind to get into more storytelling mode a little bit. Earlier this year, I was in Denver with part of our team there. We were at a bowling alley for like a team gathering, and members of our finance team were like giddy wanting to show me how they were slacking with their agents, and their agents were just automating a huge amount of the toil work that they had never really liked doing around BVAs and month-end close. And they were just like giddy that they could be bowling while a bunch of this toil work was happening while they were just having fun. And then, you know, they obviously had to review, there was a quality assessment component to it. But I don't know, I come into this very optimistic around there's a lot of knowledge work jobs that just have lots of toil. And you ask anyone, can we turn that piece of your job into a more automated process? Give you something that can streamline that piece of it so you can spend more time on judgment. And I think almost any like logical person is going to go, yes, please. That that would be incredible. You know, in social media management, posting content, in reporting, evaluating and generating, you know, board decks. Like there's so many examples in every function of toilsome work. So I don't know, I jump we have definitely across the board examples in Gusto, but anytime anyone gets to that moment of toil work being streamlined and simplified, the reaction I see is giddiness, like just just joyful giddiness they can now spend time doing higher value-add work, frankly. Sorry.
Oh no, we are seeing that at some companies there have been escalations and layoffs, and it makes sense why employees are worried about all these AI agents coming in and like 10x-ing and all that, like worrying about, well, what is my role still here? So, what have you found that you can do to make employees feel better, or what trends are you noticing among companies that do have to resort to layoffs?
Yeah. What I So, people don't love change by default. We like to feel really comfortable, and I think I think startup employees or founders I think are maybe a little bit more accustomed to change because you're you're just you're searching for product-market fit. Things are happening all the time. So AI is this exciting new thing, and and then you go to the the enterprise, and we actually just worked with a a Fortune 50 company that laid off 13,000 people and wanted to provide AI training for those people, which is honestly a really cool initiative that they were providing. And I think so it is real, there there is actual displacement happening. And I think sometimes that's because of AI, and sometimes it's masked because of AI. So it's, you know, I think there there is a little bit of that happening. But I do think that the generally the the energy around like, hey, this is a this is a new world. We can all go through our little morning of, all right, my job is different forever. I can no longer I no longer write code or I no longer do whatever that I used to do. And if I embrace it, then I can really thrive here. I can actually become 100x more valuable. And I think that that because humans still need to be in the loop for that last 1%, 5%, 10%, and the rest is, you know, done by AI, it means that that like 1% that fills an entire job is actually going to create like 100x more or 10x more or 50x more, whatever the the math ends up being based on your job. And people are willing to pay for that. And I think we see that with with sort of like the new grad hiring trends where you see some new grads join a company and they're more valuable than they've ever been because they're training senior colleagues on technology, and that's never really happened, right? Usually you you welcome a new grad in and you're teaching them how to do things. On the flip side, you have a new grad if they don't learn AI, they're like they've never been more useless, which is sad because they they're trying they're coming into an organization and they don't know the most important technology, and they also need to get ramped up on, you know, how to be a professional. And so I think there is an adoption curve, there is like change, and I think if you can paint the the good stuff, which I think there is so much of it, I think people are are open to it, and maybe or maybe they're just being forced to be open to it, but I think I think they're getting there.
Two things I'd quickly add: mindset matters. If someone wants to be successful in a bigger company, it's all about leaning into change and growing impact scale, merging multiple jobs and getting excited about that, not feeling like it's an obligation. But I'll just plug, we have Gusto Insights that is just public data we publish about the state of employers and small business in America. Anyone can check it out, but that data shows today there are companies hiring less across the US, but there's also a lot of new businesses being formed. So part of this is also folks who might have been thinking they're going to be on a career track in a bigger company deciding, frankly, they're going to now just go start their own company instead, and that's leading to a lot of new business starts.
So let's say that you're a founder or a group of maybe two or three co-founders. Given what AI can do, what do you think are now the most important early hires that they can make?
