What does it actually look like to work with hundreds of AI agents? In this episode of NEXT with John Koetsier, I chat with Steve Ancheta, founder and CEO of Zig.ai, an AI-native platform for relationship-driven sales.
Steve explains why what looks like a single AI agent to a user can actually be a swarm of hundreds of specialized agents working behind the scenes, handling copywriting, follow-ups, CRM updates, next-best actions, orchestration, and more.
We also dig into how Steve personally uses AI agents as force multipliers across his work, why he still writes important investor and customer messages himself, and why simply giving an agent access to documents isn’t enough. Steve shares his approach to building an effective agent harness, including two critical pieces many people overlook: governance and grounded truth.
We also discuss OpenClaw, NanoClaw, cloud-based agents, agent security, human-in-the-loop workflows, critical thinking, AI productivity, and the shift from reactive agents that wait for instructions to proactive agents that start doing useful work on their own.
Enjoy!
Topics include:
- Why Steve believes many beginners shouldn’t build agents yet
- How hundreds of specialized agents can operate as one system
- AI agents as force multipliers rather than human replacements
- Why human taste and connection become more important as AI improves
- How Steve uses agents for planning, operations, scheduling, and analysis
- The dangers of AI-generated documents nobody actually reads
- Agent governance, guardrails, and permissions
- Giving agents a reliable source of truth
- Why vector databases alone aren’t a magic solution
- Balancing human interaction with AI-assisted work
- The move from reactive AI agents to proactive AI agents
- Why humans should remain in the loop for external actions
Transcript: 100s of AI agents
Steve Ancheta: So this is where I think I might piss a lot of people off, but my thing is, if you don’t have any sort of AI knowledge—and when I say AI knowledge, I don’t mean novice knowledge where you have it write you an email or, you know, create you a document or something that’s very, very foundational—you probably shouldn’t build agents yet.
John: I don’t know about you, but I’m using agents more and more right now. I have two that are doing regular work for me, and that’s not counting ChatGPT and Claude and all the other things that I’m using as well. But it can get a little out of hand.
Today, we’re chatting with somebody who is used to having agent swarms up to 100, perhaps. His name is Steve Ancheta. He’s the founder and CEO at Zig. It’s an AI-native platform for relationship-driven sales. Steve, welcome. How you doing?
Steve Ancheta: Hey, pretty good. How are you doing? Thanks for having me.
John: I’m doing well. Thank you for being here. Am I talking to Steve or his agent?
Steve Ancheta: I know you’re talking to Steve right now.
John: Okay, okay. Sometimes it’s hard to tell. I mean, like, I’ll tell my agent to send an email for me, and it has standing instructions—and this is a NanoClaw agent—it has standing instructions to identify yourself as John’s AI assistant, but it doesn’t always comply.
Steve Ancheta: That’s a recurring issue with agents. They’re a little bit of a black box still. I think people don’t quite understand that just yet. But they’re getting better. They’re getting better and, you know, the models are getting better, so naturally the agents get better with the models.
John: Yeah, yeah. Maybe let’s start here. You’ve got a platform for agents and you’re—
Steve Ancheta: Sure.
John: —using a ton of agents. Where did your journey with agents start? Talk about how you got into it.
Steve Ancheta: Yeah. So, man, that’s actually a funny story. Quite a few years ago—well, I’ll go back actually really early to a tweet back in, I want to say it was 2022. It was a tweet by a gentleman by the name of Sam Altman, who a lot of people know, and he said—yeah, yeah, exactly—and he said, you know, “OpenAI is excited to announce their product ChatGPT. It’s live. Go try it now.”
Actually, I think it was before that. I think it was when he said to get on the beta list, and I did. And so you were on this wait list, and I think I got access to it maybe a few days earlier than the public release. It wasn’t too much earlier, but, you know, I started using it.
I still have screenshots on my phone of the first version of it. And I don’t think we ever thought that we would be here, and I don’t think we ever thought that we would get to probably where we’ll be in the next five years. But when the agentic kind of transition happened, it was very much in its infancy.
So I want to say this was 2024. I was running a company, and I had been using ChatGPT for quite a few years at that point, had gotten pretty good at it. I knew it was the future of technology and, you know, another company who’s very well known, n8n, came to the market, and they made a pivot into—
John: I—
Steve Ancheta: —workflow—
John: n8n, but go ahead.
Steve Ancheta: Yeah, yeah. They made a pivot into workflow automation, and it was kind of the first engagement you could have with, like, building an agent. Now, granted, the agent could do one piece of one thing, and you would have to build multiple into a single workflow to get it done. And, I mean, it was rough.
I think I still have some of my old n8n workflows saved probably in my workspace. I don’t use it anymore. Haven’t used it in a very long time. But it was a mess. It was a mess, but it was just so interesting to have, like, man, there’s a computer essentially that can do things for me if I just command it to do so correctly.
And yeah, so that was, like, the early part of my journey with agents. I remember saying—my company was doing rather well—and I remember saying, like, “All right, I got two options. I can either sell this thing earlier than I want to and go full, you know, head-deep into this industry that I’m finding a passion for, or I can wait and possibly miss the boat.”
And so I sold my company. I had the sharks swimming at the time already, and so I just reached out to some of those sharks that were swimming. I said, “Hey, look, this is probably your one chance. And so, you know, I don’t know if I’m crazy or what it is, but if you want to buy my company, this is the time to do it because I’m gonna go on to do something else.”
And so that’s kind of what it was. In the early days, it was really messy. Now it’s gotten a lot easier.
John: There’s a word I didn’t hear in the story of your genesis with agents, and that was OpenClaw.
Steve Ancheta: Yeah. I mean, when OpenClaw came to market, I did use it. I prefer not to knock any one over the other. I think Hermes—or Her… Yeah, I like Hermes a little bit better personally. But OpenClaw is great. I thought OpenClaw was probably one of the coolest things to come out.
The issue with OpenClaw—a lot of people who are really passionate about agents jumped into it. The funny thing is, living in the Bay Area, which is one of the hubs of innovation in the United States, there was an event in the lounge of my apartment building that was, like, an OpenClaw hackathon.
It was a two-day hackathon, and I went, and it was really, really cool. The issue with OpenClaw, though, is in the early times—again, I haven’t tested OpenClaw in quite some time—but security was always a big issue, right?
John: It was almost nonexistent for OpenClaw. I mean, you could do—
Steve Ancheta: Yeah.
John: —initially. That’s changed a little bit, but—
Steve Ancheta: Yeah.
John: Yes.
Steve Ancheta: Yeah. As most AI companies do, they mature.
But yeah, so security was an issue. The other issue is if, you know, your computer had to go through a restart or went offline or went to sleep. A funny story was I had a sales rep—a VP of sales, actually—who I was really good friends with. I met him at a conference, and I was showing him, like, the really early versions of Zig, and he loved it because, at that time, early on, we made it cloud-based.
And he said one of his biggest issues with OpenClaw—because he had built pretty much what we had built in Zig in a personal version in OpenClaw, more or less a few of the things that we had that he could do—he was like, “The problem I had is I was on a flight and my computer, or my Mac Mini back at home, gave me an alert that my Google API key had knocked offline, and there was nothing I could do because my wife’s not gonna know how to do it for me.”
So really, I was dead in the water. He was like, “And I wasn’t gonna be back home for a week and a half.” So that was kind of some of the things that… And so, yeah, no, I think it was awesome, though.
John: I have mine—
Steve Ancheta: Sure.
John: —set up that if it has a problem, it’ll restart, and everything is authorized to restart. Doesn’t use OpenClaw. I didn’t want to go OpenClaw. I went NanoClaw, which is a more secure kind of… But yeah, I mean, if your agent goes offline, you can’t get there for a week and a half.
I mean, the cloud works, right?
Steve Ancheta: Yeah. Yeah, the cloud works. I mean, it’s—and everything’s becoming more and more advanced, right? Like now we’re—it’s not even, you know, we’re not even in the era of cloud versus personal machine, because even though I have my normal cloud version of, let’s say, Zig, I also have a locally hosted version that I do a lot of iteration on whenever I have some ideas that I kind of want to toy with without touching our back end.
Plus, my engineers would never let me touch our back end, for good reason. I wouldn’t want to. And so, I have autonomy on the front end. But it’s not even becoming that anymore now. I mean, what our company is doing is, like, you can containerize it, and you can host it inside of a company’s—your customer’s—environment.
And so it’s becoming this kind of lift-and-shift technology, which is awesome. I think it’s great. And if you think about it, I know this is a really weird analogy, and actually this is the first time I’m ever making it, so if it’s bad, it’s gonna be really bad. But I compare it to, like, the house phone versus the cell phone.
Think about the years of separation between the house phone being invented versus the cell phone being invented. Now think about the difference between an agent having to be locally hosted and an agent being anywhere, right? Whether it’s via MCP or what have you, being anywhere you are.
Think of, like, that distance closure, right? Same kind of technology shift between those two, but months. You know, what was decades before became months, essentially. And yeah.
John: Not a bad analogy. I think I like that analogy, and it’s super interesting. Like, how long did you keep a landline, right? We got rid of our landline pretty quick after we got cell phones. There was just no use case on that that really mattered. I mean, obviously there’s the emergency cell network is down, but we really haven’t had that problem.
So, let’s talk agents. You’re building a platform that utilizes agent swarms. Maybe give us a sense of how you use agents, and then we’ll go into a little bit more. And how many agents do you have, and how do you use them?
Steve Ancheta: Yeah. So, it’s kind of funny, and I’ll talk about the product and I’ll talk about how I use agents personally as well, because I think they’re two different conversations. The product itself is hundreds of agents. Agents operate best when they have a very defined task that you can work around and debug and, you know, iterate and reiterate.
And so it’s so funny. There’s a big difference between communication to your customer, like marketing—there’s the marketing jargon and then the actual reality, I think, in AI, and it’s happening across the board. To the customer, it looks like one continuous agent, right?
It’s Zig. But behind the scenes, it’s swarms. It’s hundreds that are handling these tasks and doing handoffs, and there’s orchestration and assignment and, you know, a new agent creation capability and things like that. So, in reality, it’s hundreds, but in the way that you speak about it, it’s one, because once you start saying “swarms,” it starts making people’s heads spin a little bit, I think.
And then the way that I use agents personally—and in the… Well, actually, to close that loop, in the product we have agents that are specific in copywriting, and they write, you know, whether it’s a follow-up email versus a cold email versus a LinkedIn message when a sales rep is trying to essentially get into a prospect.
There are agents that are specific in follow-ups, right? Next best actions, CRM updating. Each of those has its own functionality and surface area that it controls, and there’s the orchestration layers above it that are invisible to the user, but that’s always the most important because it has to direct traffic, right?
And so it becomes almost like hops. My background’s in networks, so you’ll hear a lot of network and telecommunications analogies. So, you know, it ends up becoming like a network. There’s hops. One agent orchestrates, it realizes that it has to go to another, and then that agent categorizes it into its sub-functions and then drops it into its sub-agents, and then the task gets done.
From the user standpoint, the task just gets done. That’s all they care about, right?
John: Yeah, yeah.
Steve Ancheta: And then the way that I use agents personally is, man, I think I use agents to do almost anything and everything that I do, except one thing. There’s one thing that I’ve gotten away from using agents on, and I think it’s kind of funny. I have gotten away from using agents when I do investor updates, when I do customer updates, and when I do one-to-one customer emails a little bit.
Unless it’s a very brief follow-up that’s a reply to another reply. But if it’s something that I’m coming to either a customer, an investor, whoever it is, I handwrite it. And the reason why is because it doesn’t matter how good the copywriting agent is. I mean, we have an entire team internally that is focused on copywriting of our agents because our thesis when we started was our copywriting needs to be better than anything that’s on the market.
That’s the only way that we’re gonna win in this kind of world of noise, and specifically for that category. And so, our agents write great emails, but there’s still this thing of, like, you can tell when a human is being put through the messaging.
John: Make mistakes on purpose or just leave the mistakes so that—
Steve Ancheta: I leave the mistakes. I swear I do. Like, if I misuse a there or their, right? Or if, because I’m typing so fast, I hit an F instead of a D or whatever, I’ll just leave it.
John: Thinks it knows how to talk Eng—
Steve Ancheta: Oh man.
John: You know.
Steve Ancheta: It drives me nuts.
John: We’ll ignore that because I want some idiosyncratic communication. I want it to be me and not just, you know, some pabulum, right? So—
Steve Ancheta: Sure.
John: Yeah. Okay, that makes sense. That makes sense. But you’re using agents for pretty much everything else, and let’s get into what those things are and what those are useful for.
For me, I mean, when I look at agents in general, they’re force multipliers. The Steve Jobs phrase, “a computer is a bicycle for the brain,” comes to mind, right? It’s just, I literally have employees. I literally have staff, and I can hand off tasks and get them done. And yeah, I’ve gotta sometimes communicate between them, and sometimes I’m the conduit between them.
And yeah, sometimes they do something wrong and I have to correct them and everything like that. It kind of really is like having staff, but it’s a force multiplier, and it allows you to do 5X, 10X, in some cases 100X what you could personally do. It’s literally insane.
Steve Ancheta: Yeah. Yeah, I think you’re 100% right. They are force multipliers. I’ve always had that thesis. We play in a very busy space in the sales go-to-market space. And there were a lot of companies that came over the last few years that were, like, trying to go into the “replace humans with AI” and, you know, you can have these AI SDRs that are gonna just do everything.
And now every single one of them is rolling that messaging back because I think they’re realizing very quickly that the human element—and this has been my thesis since the beginning—is that as AI becomes more of an adopted technology, you’ll hear me say this all the time, the human element becomes equally as important.
And I think that’s something that everybody didn’t think was necessary. But it’s very weird. I’ve been in sales my entire life, sales and operations, so I’ve managed companies and sales teams, and I was a sales engineer and, you know, I’ve held every position inside of a sales organization.
And one thing I can tell you is that the reason that sales is my favorite thing in the world is because it is very heavily tied to human psychology, and humans are connection creatures to their core. And if there’s a lack of human connection, then it becomes very hard to get things to a place where they feel comfortable.
And so, I think that as we built our company—and I see a lot of companies doing this now, which I’m happy for—they’re trying to use AI to be a force multiplier to essentially make the human aspect come out more and more. And I think you’re seeing that a lot with, I think, one of the most important human aspects, which is taste.
John: Mm-hmm.
Steve Ancheta: Someone has had their hand in, their DNA in, something. Yeah, and so I think that’s only gonna get more important.
John: So that makes sense. 100% agree. Follow along with all that and agree with that. What are you using agents for as force multipliers? What do they do for you?
Steve Ancheta: My favorite, honestly—so, one, number one is planning and documenting, right? I think that’s a huge one for me. I think that’s a big one for me.
John: Literally awful tasks, like—
Steve Ancheta: Horrible.
John: Like—
Steve Ancheta: Tasks. Yeah, but you also have to be extremely careful with document creation. And the reason I say that isn’t for legality or things.
I mean, that goes without saying. But you have to be very careful with document creation because I’m starting to see a lot of document creation where somebody’s using an agent to create a document, but they’re not actually looking at the document, so they don’t really know what’s inside of it. And that’s my one fear with AI, and this is as somebody who runs an AI-native company for a living in Silicon Valley. I mean, I’m in the thick of it.
I’m worried about critical thinking a little bit because critical thinking is a very human element. Take something as simple as document creation. If you’re not gonna review your documents, then that means there was no critical thinking put into the document other than what you spewed out on a, you know, one- or two-kind-of run.
John: The 20-page AI-generated nonsense report from some asshole who then we have to spend half an hour reading it or refuse to. Absolutely. You can create a lot of garbage and send around a lot of garbage and waste a lot of people’s time, and if you’re in a position of power in an organization, that is exactly what you do not want to do. But what are your agents doing for you?
Steve Ancheta: Yep. So, outside of things like document creation, planning week to week, quarter to quarter, month to month, kind of based on ongoing strategy shifts that I have. So I have agents that, you know, I build skills within my agents, and I use different formats for different things.
Like, if it’s something as simple as document creation, I use, let’s say, Claude Cowork. But then if I’m doing something that’s much deeper, then I move to Code or Codex and, you know, I use that for more of the complex things. I feel like the output’s a lot better.
And so I have agents who are constantly, you know, overlooking company operations, overlooking things like projections, and then also what I use it for is I use it to see if I’m missing things a lot.
So, “Hey, I’m projecting X over next quarter. Is there anything that you’re seeing across the company that is actually—you know, we’re not gonna hit this deadline because this and this and this is happening inside of Notion, or it’s happening inside of, you know, you’re seeing it in a Google Drive somewhere, that something’s slowing us down that I’m not privy to?”
And so I have agents that are constantly scanning and surfacing those types of things and giving me reports. I also have agents that keep—and this is outside of just using Zig. This is more of just an operational-type aspect. I use it to make sure that my days are optimized, right?
That the way that my days are scheduled are optimal and that if there’s things that I need to do, I’ll do a brain dump and say, “Hey, these need to be plugged in where they’re the most necessary, or by priority, and then also where they’re the most necessary.” And so it’s almost like a chief of staff/executive assistant.
But the funny thing is, one of the hottest jobs in the Valley right now is the human chief of staff. And so it’s just really weird how that works, right? AI does what people think, like, a chief of staff or an executive assistant can do.
But obviously, there’s still things that are missing because the human chief of staff kind of job demand is going up and up over the last few months.
John: Yeah, yeah. Okay, so you got multiple agents running. There’s challenges sometimes when you have many agents running. Are they working on the same thing? Are they duplicating work? Are they having an iterative, recursive conversation amongst themselves and just burning tokens and not actually producing anything good? How do you make sure agents are doing what they’re supposed to be doing, not going off the reservation?
Steve Ancheta: Yeah, so this is where I think I might piss a lot of people off, but my thing is, if you don’t have any sort of AI knowledge—and when I say AI knowledge, I don’t mean novice knowledge where you have it write you an email or, you know, create you a document or something that’s very, very foundational—you probably shouldn’t build agents yet.
I think there’s still complexities that maybe those folks may not understand, and you can get yourself into things where you’re now, you know, communicating information that is incorrect because of kind of the parameters that you built it around.
So building, like, an agent harness is, I think, one of the most important things anytime you’re gonna build multiple agents. There’s always a harness, a grounded truth that needs to be built. And it doesn’t happen overnight. It’s something that—it’s a surface area that you start really, really small, and then you give it a little bit more, a little bit more, a little bit more, and that’s how it builds context.
And so that’s why even with Zig, I tell our customers, like, when they ask, “Well, are you guys a—you know, is it… I know it’s an AI-native tool,” especially because the markets we sell into are still very much behind on their AI journey, right? As an organization or as a company, or their employees aren’t there from a technical acumen standpoint yet.
So we make sure to make it make sense to them. So they ask, you know, “Is it a—I know that it’s gonna make my reps faster, and they’re gonna be able to now kind of manage these AI employees, it sounds like, and this and that. But are you guys”—and this is usually in our early conversations—”a productivity tool?”
And I’m like, “No. We’re actually a data company,” because the foundational data is our product, right? The foundational data and the agents that use that foundational data as truth is the product itself. The UX—
John: Mm-hmm.
Steve Ancheta: We have, you know, MCPs inside of most models. We have a single OS that you can work out of. We can lift and shift the way that it looks. But really, the product is: is the data correct and accurate, and is it mapping correctly? I think that’s the biggest thing.
I always give this analogy of, like, if you just give your agents very minimal context, it’s like putting them in a pitch-black room and not telling them what’s inside.
If you give them some context, being, let’s say, documents, which is what a lot of people default to, it’s like putting them in a dark room, giving the agent—and pointing at a maze that is dimly lit and saying, “Okay, go find the end.”
If you give them a true foundational layer, it’s giving them GPS coordinates of where they are, an aerial view of how to get to the end, and a flashlight.
Like, there’s different levels of building that foundational layer, and I think that’s where I feel like novices aren’t there yet to where I would tell a novice, like, “Go build yourself agents.” You’re just not gonna get the success. You’re gonna get frustrated. You’re gonna abandon it.
I’ve seen it. I’ve seen it time and time again.
John: We see a lot of that, especially early on in the OpenClaw revolution where people were getting OpenClaw up and running, then what? Or—
Steve Ancheta: Yep.
John: —trying to get it up and running and just flailing and—
Steve Ancheta: Yeah.
John: —giving up on there. I don’t know exactly what you mean when you say giving it a harness, but that’s a common term, obviously—
Steve Ancheta: Sure.
John: —in AI, and it’s basically, in my understanding, here’s what you do, here’s what you focus on, here’s what you…
Like, for my agents, what I’ve done is that this is what you do. This is your job. You know, I’ve told them explicitly what their job is, what they need to do, what they need to do for it, who they talk to, what resources they access, and all that stuff. So there’s explicit training there. Is that what you mean by a harness?
Steve Ancheta: I think you’re on the right track. I think there’s two layers there that I didn’t hear mentioned. One of them is governance: what surface area they’re not allowed to touch versus what they are allowed to touch. That’s very important.
John: Mm-hmm.
Steve Ancheta: If you don’t give them those guardrails, then, you know, you could run into some hot water.
That’s the second one. And then also, what’s their truth? Like, what truth do they have to anchor decisions in? And maybe for a personal agent, it’s not as important.
John: Mm-hmm.
Steve Ancheta: But for an organization where there’s many different pieces of truth and they’re all fragmented and don’t talk to each other, it’s very, very important.
And that’s what we do. That’s where our focus is.
John: Mm-hmm.
Steve Ancheta: You need to build that truth. And there’s a couple different ways to build it. In the early days, it was vector databases. Vector databases are an absolute shit show at times. And not because the technology’s bad, but it’s more because, you know, people expect it to be kind of this foolproof, like, I could do a data dump and then my agents are gonna know exactly how to pull the right data every single time.
That doesn’t work.
John: Mm-hmm.
Steve Ancheta: And so making sure to architect its truth as well as its governance are the two most important parts of a harness.
John: That makes a ton of sense. Absolutely agree with that. Cool. So as you move forward and as you’re using tons of agents, how does it change how you work? I mean, obviously it makes us more productive. There’s a force multiplier. You can send an agent off to do a job, you can send an agent off to do a job.
I often feel sometimes when I’m doing things that if I haven’t given this agent a job or I don’t have Claude doing this, I’m wasting time, right? Because I can have them busy while I’m doing something else, and then there’s a sort of continuous partial attention thing that goes on, which is really weird and maybe isn’t great for all times, but can be super productive. How do you work with that?
Steve Ancheta: Yeah, I split my day up, right? I try to have equal parts human interaction, whether it’s customers, investors. I like having ideas because that’s the other thing, is agents also need direction.
How many times—or maybe this has never happened to you, John—but have you ever sat in front of Claude and been like, “I don’t really know what I wanna tell—like, what I need you to do right now”?
It happens all the time. I mean, it happens all the time where somebody will sit and just, like, “I don’t have any input to give you so that I can get whatever output I think I’m looking for.” But speaking to other humans sparks those types of ideas.
So I try to have equal parts human interaction as I do Claude interaction or GPT interaction or any of my agent interactions.
And then I like to spend a little bit of time out in the wild, you know. I get a lot of inspiration from architecture and art and landscapes and things like that, things that have nothing to do with technology. And then they give me, like, these ideas around technology because it gives you this head-clearing thing.
Like, you can go in with an agent and go back and forth for hours, and then you’ve lost hours of time. So it can be equally productive as well as counter—you know, it can also counteract itself at times. So I think it’s really important to structure your day and week, which is why I dedicate time to deep work versus idea-based work versus, you know, customer-based work, things like that.
John: That makes sense. It is a different way of working, and there is this AI generation thing where we’re working with AI, we’re working with agents. Being productive in that world is different than it was to be productive a decade ago or—
Steve Ancheta: Sure.
John: —ago, and I’m sure there’s things that we gain from that.
I know there’s things we gain from that. I also know that there’s things that we lose from that. So having that level of balance and saying, “I’m gonna be in that world, but I’m also gonna be in a different world”—the physical world, the real world, the relational world, the world where I take more time to just think deeply and focus without dings and this agent coming in.
I think that makes a ton of sense. This has been super interesting, Steve. It’s a wild, wacky world, and I look forward to seeing what agents become for all of us.
Steve Ancheta: Yeah, me too. I think the era that we’re entering right now, that we’re seeing—we’re building it, we’re one of the companies building it—is agent proactivity versus reactivity. Agents, you’re seeing it, you’re seeing startups pop up that are doing this very well. We’ve been working at that and releasing it little by little over the last few months, is agents that’ll just do things.
So, you know, inside of Zig, if it’s inside of the customer ecosystem, agents can work in a proactive fashion. The only thing that our agents aren’t able to do right now, and we did it by design, is they’re not able to do anything to the outside world. So my customer can’t reach out to their customer without the human-in-the-loop step piece to it for approval.
But things like, you know, pulling leads or creating campaigns or things like that, it’ll just do it, and it’ll tee it up. And creating follow-ups it’ll do, and it’ll just tee it up for the human to just basically say, “Yep, that looks good. Go ahead,” instead of them having to ask it to do so.
And so I think we’re gonna see more of that in other industries and in other, you know, parts of technology. And that’s what I’m really excited about because I think that once folks feel like, you know, the outcome’s important or the goal’s important, you’re gonna see industries change very quickly.
You’re gonna see job functions change. You’re gonna see the way… I think it’s gonna be a really, really cool thing to see. And I think that there’s gonna be a lot of people that are gonna do very, very well over the next few years because of adopting and really iterating with the technology.
John: I agree, and it’s pretty interesting. I’ve got some agents that do that myself, and right after this call, I’m gonna go and see what they’ve done. You know, it’s funny because we tell them what to do, and then they give us back stuff, and it’s on our task list. It’s like the agent’s telling us what to do now, building my task list.
Well, thank you so much for this time. I really appreciate it.
Steve Ancheta: Hey, no worries. Thanks, John. Really appreciate it.