The Agent-Native Company

Talk transcript

1. 0:01

   Hi, my name is Rick from Agentuity, and today I'm gonna talk to you about agent-native companies or AI-native companies. What that means, you hear a lot of that right now.

2. 0:10

   What does it mean and, uh, what are some of the core attributes and things like that? So, so imagine, if you will, you walk into the office, pass your coworkers, smile away with them.

3. 0:21

   You sit at your desk, you sit down and you check not your email, but the status of what all your agent coworkers did over the, the night before, did last night.

4. 0:33

   Uh, they're pulling you into things that need your attention, things that they got done or things that they think need approval. Um, and so instantly you get brought into their world on, "Hey, I need to go do and execute these things because that's part of my job."

5. 0:48

   So for a lot of us, I don't think it's actually hard to imagine, especially for us that are doing a lot of these AI things right now. Uh, certainly at our startup at Agentuity, uh, we're feeling exactly that.

6. 0:59

   During a team meeting, it's really common to hear things like, "Well, just have Devin go do that." "Oh, that's a good point. We need to go update that. I went ahead and pinged Devin."

7. 1:08

   And then by the time we're done with the stand up, I've already got a pull request that we have to review. Or things like, "Hey, we should tweak this channel.

8. 1:15

   Um, let's go ping the content marketing agent and it'll, it'll adjust it." That kind of thing. So, um, those things are, are common more and more and more, especially at Agentuity, and we think that for a- an agent-native or an AI-native company, some of those things are just gonna be the norm.

9. 1:32

   I mean, even as an example, I was on X recently and a startup posted a job opening for an Agent manager. That's a manager that manages AI workers. Pretty interesting, right?

10. 1:44

    So we're entering a world where managing agents can be someone's full-time job, literally.

11. 1:51

    So in this talk, I wanna explore that world, agent-native, AI-native companies where AI just isn't an add-on or just some, just some task that somebody uses, but it's central to how work gets done.

12. 2:05

    So we're gonna walk through a few things. We're gonna walk through AI-enhanced versus agent-native. There's a difference, or AI-native. Um, what makes a company agentic, um, the typical agentic workday, and, uh, even just like rethink hiring.

13. 2:20

    Like maybe we need to think about hiring a little different. So that's what we're gonna cover in this presentation. So let's just jump right into it. The first thing is AI-enhanced versus AI-native or agent-native.

14. 2:32

    So what does that mean, uh, anyway? So what is a native company that uses AI? What does that mean? So hopefully you're not one of those people that think that an agent is only a chatbot. [chuckles]

15. 2:41

    If you are, then you're probably thinking, "Well, why would you... What does a chat-native company mean?" So if you are one of those, then we probably need to make another video, um, to talk through that, but my, my intent is not to talk and convince you what an agent is.

16. 2:55

    Um, it's not just a chatbot. [chuckles] But in a nutshell, what we mean when we say agent-native, um, company is a company built from the ground up with AI agents at the core of everything to augment human productivity and intelligence.

17. 3:13

    Um, I'd, I'd also additionally add that, um, these companies build AI into the foundation of their product, into the foundation of their operations, and into the foundation of their culture.

18. 3:26

    Another way to look at it is AI is not a fancy side feature that's just a bolt-on, um, to existing operation and culture. It's an engine that moves the product operations and culture.

19. 3:37

    It's the thing that moves it forward. Everybody reaches for that. Everybody relies on it, uh, to do their job. So sometimes it's helpful to think about this in the opposite to contrast.

20. 3:46

    So for these companies, if you remove agents, like just for us, for our company, Agentuity, if you just remove agents out of our workflow, um, guess what happens? Our employees aren't able to get as much done.

21. 4:00

    They're just not. The, the lot of the mundane things that we do, unsatisfactory work that we do, we have to now do, and, um, it, it doesn't, it's not as fulfilling.

22. 4:10

    I mean, j- just even for us, we have an agent that, uh, writes our changelog and our documentation. So after all this stuff happens in GitHub, all this code gets written, all these things get deployed and released, um, we have to create a changelog.

23. 4:22

    We have to create documentation. We don't do that. Uh, we have an agent do that. And so if we, if you were to remove that from us, we would, we'd be very sad.

24. 4:30

    All of us as engineer, software engineers would be very sad if that got removed from us. Um, another one is if you think about it, removing AI or agents from a agent-native company, um,

25. 4:42

    makes us not move as fast. The costs go up, pro- productivity goes down. Removing AI and agents, um, from the products that we're building, products start to feel old, not as interesting, unintelligent, not as useful.

26. 4:55

    They certainly don't help the customer 10X themselves. We're wanting to 10X or 100X ourselves. How do we do that with our customers and our product, right? Um, so then all of a sudden now we're not as good as the competitors.

27. 5:06

    The list goes on and on and on, but that's the counterpoint to it. A way to state it is merely an AI-enhanced business, one that uses AI here and there, chats with it every once in a while, gets a document every once in a while, maybe that helps with some efficiency goals.

28. 5:21

    It's good. And they would still function without AI. That's the key part. If you removed it from it, it would just... it would still function, it would just not be as efficient.

29. 5:31

    And that's a car with driver assist. That's really what we're talking about. An agent-native business is different. It is a car on autopilot, directed by humans of course, but it's still autopilot.

30. 5:40

    It's a different kind of car. Its signature is every employee is focused on the higher level navigation task and the success of the company and the product, while the routine and mundane tasks, micro-decisions, they get offloaded.

31. 5:55

    That's a signature part of it. So the thesis that we have is that an AI-native or agent-native model isn't just a tech trend. It's, uh, it's redefining how we build teams, how we design workflows, even what roles we hire for.

32. 6:10

    I don't think it's any different than back in the Industrial Revolution and during the car age and some of those things, how everything got transformed. I think it's just like that.

33. 6:19

    So now moving on, uh, what makes a company agent-native? Now, obviously we're on the early days, very early days of course, but here's a few defining things that make an agent-native company.

34. 6:32

    AI is at the center of everything. Well, you say, "Rick, yeah, no duh." I mean, just listen though, it's, it's not confined to just one team or a feature or one aspect of a company's culture.

35. 6:42

    It's everywhere and it's everything. Think of product, think of customer support, think of ops. In an AI-native company, the expectation is each of these departments have agents doing some of the routine and key daily work all the time.

36. 6:58

    The departments would have agent interfaces, hand off to integrate with other departments. I mean, this is obvious, right? In order to be efficient, you need this. It should be obvious.

37. 7:09

    But again, think of the opposite. If you turn that off, each department, uh, if you turn, turn off all their agents, um, you know, you got a human scramble trying to coordinate manually, feeling a loss of productivity because you had this thing, you had this way to automate things and to work faster and to work more, and

38. 7:25

    now you don't have it anymore. Um, man, that's a big, big, big, uh, loss, uh, for, for us. Like just thinking about that is painful right now. We have six people in our company, seven people, and, um, removing that would be, would be bad. [laughs]

39. 7:40

    Um, the other thing is with an AI or an agent-native company, people are no longer just like cogs in a machine. That's another way to look at it. They're more like conductors.

40. 7:50

    Now, if you think about that statement for a minute, a, a lot of people make that statement. It, it's not a novel statement. It's a, it is an important statement though.

41. 7:58

    They're conductors. You'll realize that the hiring profile and the company org chart will need to be different if that's who you're hiring, if that's what you need. Flatter, leaner is a key attribute.

42. 8:12

    It starts to become... It can be a, a key attribute. Middle management layers shrink because a lot of that coordination can be handled and executed by intelligent systems. I mean, even in my own experience, we'll have a deep dive product discussion in the morning with the team.

43. 8:27

    By the end of the day, we have detailed requirements, um, to work on something. We already have had agents start to build it. It's already helped us with the messaging and the copy and some of the documentation, and that's just within, you know, a day.

44. 8:40

    And so we've had this culture now like, well, you, well, yeah, this is a great idea. Let's prototype it. And we prototype it in a few hours, and we get some more refinement, and agents help with that a lot.

45. 8:49

    They help us do more prototypes and more learning and more testing that way. So that's why we think at least our org chart in the near future as we scale is gonna g- look less like a pyramid and more like a network of humans and AI together.

46. 9:05

    Another attribute is experimentation and iterative culture in the DNA. So now I know, I know, I know, I know this is a core value in, in the tech startup world.

47. 9:15

    Core value, core innovation is that experimentation, iterative approach to things. So it can be cliché, but if you think about it in context of an AI-native company, with all the innovations we have now, the ultimate realization of this is really possible.

48. 9:32

    When AI is doing the routine work and AI is helping us with the prototypes, we can really focus on what matters. I think it's, I think it's powerful and I think it's awesome.

49. 9:42

    And it also compounds. When you start thinking about agents working on things in the company learn and improve, it really is useful. I mean, just take Cognition's Devin. We use it.

50. 9:52

    We started using it a few months ago. It was good. We had to do some things. And now over time it's learned so much, it's documented so much of our code, it knows how we handle certain things.

51. 10:03

    It's, it's a superpower at this point. So all of these attributes, AI at the core, humans as orchestrators, uh, rapid experimentation, self-learning, agent evolution, they combine to create an organization that looks and feels totally different from a traditional company.

52. 10:21

    It's not just a little more efficient. It's operating in a totally different model. And so like if you're an MBA and you're like, "I wanna... I learned this approach at Berkeley, you know, and this is here.

53. 10:31

    Here's my template," it's not gonna fit. It's a whole other model we're, that we're talking about.

54. 10:38

    Now let's talk about, uh, the agent-native typical workday. What does that look like anyway? And I think we have a lot of ideas on what it can look like in the future, but really let's just ask the question, well, what's new?

55. 10:51

    And definitely one of the things that's new is overseeing what AI is doing or what it has done. That's certainly something that is, um, an everyday part of our lives now that definitely didn't exist just a few years ago.

56. 11:04

    And so even just me, for example, I usually start my day, uh, with a built-up log of things that I have to get done, and usually I have certain theme days, so I try to put s- put certain tasks on certain days.

57. 11:17

    And sometimes they're mundane tasks, sometimes they're product ideas that need to be thought through. So I found that like in the morning, a lot of times I'm out for a walk, I'm driving the car, and I'll actually chat with, I'll talk to ChatGPT and I'll use o3 or deep research, um, some deep, um, reasoning stuff, and, um,

58. 11:36

    just kick off bigger thinking things. Like, "Hey, this is what we're thinking of. Here's a document. Here's a link. This is what we said. Here's a conversation. Here's a Granola transcript."

59. 11:44

    And then, so by the time I actually get to that to-do for that day, I've already got a bunch of thinking around it. It might have kicked off Devin and created a couple PRs, and that's definitely the case when there's, uh, things like bugs and doc issues and other things that have amounted from users that get put

60. 12:00

    into Linear. Devin's already working on it. Devin's already got PRs. So it's very, very helpful just to get things moving. A lot of times, you know, you've got like the, the molasses around your, your legs and you're like, "Oh, I just gotta get moving before I start get, uh, getting up to speed."

61. 12:15

    And so this kind of helps with the ball moving a lot. So certainly that's, that's part of the beginning of the day for me.

62. 12:22

    And I, I think essentially what we found right now is the morning time is check out what agents did, what they need to do, kick off the things that they need to do, and then come around, around, I don't know, around lunchtime, review everything.

63. 12:36

    You're gonna have a bunch of PRs, you're gonna run extra collateral, maybe some emails. We got some email agents that do things. Um, and so that's one example. Big picture, I think every employee can become a lead manager type and, and, and I don't mean necessarily a people manager, but of their AI agent counterparts that are responsible

64. 12:58

    for the jobs the person is hired to do.

65. 13:01

    I mean, we have content marketing swarms that we use multiple agents that, that I'm orchestrating and telling the agent, "Hey, you know, the copy's over here. Optimize it. Figure out the best time for social posting."

66. 13:14

    And all this happens and automatically schedules with Typefully we use. Um, so it's a, it's leveraging expertise of the human, but with async workloads, kind of just doing it all so that when you get to it, it's all ready to go.

67. 13:27

    Now, back to my earlier point, I think this leads to a much flatter team structure and probably different titles, honestly. Titles that combine domain expertise with AI know-how. I mean, we even have that in our agency, right?

68. 13:40

    That's not true, where it's always gonna be the customer support. I think so. It might be AI engineer or AI, um, AI customer lead or something like that. I mean, even our own industry, it's AI engineer is now you hear that everywhere, and I think that's gonna be across the board, across everywhere in the organization.

69. 14:00

    So that leads us to the final point, which is rethinking the hiring process and who you're hiring. 'Cause when you start thinking this way, you start thinking not just the hiring process and operation, but who you're hiring needs to be a little different.

70. 14:16

    You know, curiosity, adaptability, those are things that are in high demand, right, for creative people, for leadership roles. So we know that. But essentially what we're saying is for an agent-native company, that is required.

71. 14:30

    We, we, we have to have that right now in order to hire somebody because they're not gonna be able to use AI, uh, the way that we need them to.

72. 14:40

    So, so then maybe another way to say it is AI fluency becomes a must-have. You know, to reference, uh, the old world, you know, you would never, ever, ever hire somebody in an office job if they've never used a word processor in their life, right?

73. 14:55

    They don't know how to use a keyboard, you would never hire them. It's just expected. You wouldn't even probably put it on your job requirement. It's just expected that they know how to use a keyboard.

74. 15:03

    Now, if you went 120 years ago, that might be a debatable thing. I don't know. Is it really needed? Do they really need to use a keyboard? I don't know.

75. 15:11

    It's, uh, it's just type- typewriter. As long as they have nice hand- uh, penmanship. [laughs] Um, but AI fluency is actually a really big deal for us, and I think it's just gonna keep being that way.

76. 15:23

    It's not really hard to imagine. If the expectation is a flatter org structure where each employee is efficient at directly guiding agents for task at hand,

77. 15:34

    u- and utilizing the person's expertise through those agents, then you're gonna hone in on the ability to do that. You're going to, in your hiring process, wanna find out if that person has that ability to do that.

78. 15:47

    So it's just gonna be something that happens, I, I think. I mean, even in our current interviews, we're very, very skeptical. If someone doesn't use AI and not familiar with it, it's just immediate warning flags.

79. 16:00

    Um, now maybe they haven't had the opportunity to use it and stuff. Obviously there's, there's scenarios 'cause we're early on in this that, you know, we have to be a little understanding.

80. 16:08

    Um, but it does make us ask the question, does this person have the ability to guide AI agents and learn the little, uh, tips and tricks and tweaks and, and all that kind of stuff?

81. 16:22

    And, and e- even on a bigger picture, you know, a, a reas- a big reason why you would hire a VP, for example, you bring in a VP, is for their network.

82. 16:30

    You hire them for their experience and the quality of people that they would bring in to fill out their team to do the jobs that they need to do.

83. 16:39

    So in this new type of company, the spotlight gets really put on the, like, does this, does this guy know how to use agents? Does this guy-- Can he bring in people that know how to use agents?

84. 16:51

    Um, do they have AI fluency? And so it becomes a really big, important thing. And then if you kind of just think through the, the, the life cycle of that, okay, let's say you hire that person, then the onboarding, and then they hire some other people to build out their team.

85. 17:04

    The onboarding is probably different too, right? When you hire that, the intent being agent-native, then, uh, guess what? Agentic tools and systems that need to be put in place for them to be successful is really important.

86. 17:17

    So then it doesn't seem unreasonable to attach an engineer to that team to make sure that their agents are up and running and build out. Um, it's the same thing with, uh, like new hires.

87. 17:28

    You probably would expect them the first few weeks to solely be focused on getting their agent set up to do their job. Makes sense, right? So in conclusion, just to wrap it up,

88. 17:43

    we're going through a profound shift. We don't know exactly where we're gonna end up. You know, the ceiling is being raised so high, we don't even fully understand, I think, how large this AI world and the economy will go.

89. 17:54

    I think it's a lot like the car industry. We didn't realize how big the ceiling was being lifted in the, in the economy when the car came along, and I think that's the case with this, this AI stuff.

90. 18:06

    So I think AI agents will be deeply embedded in every aspect of business, and that means rethinking roles, it means rethinking skills, and I even think it means rethinking culture and operations.

91. 18:20

    The move isn't merely just businesses that use AI as a tool. That's not what we're talking about. We're, we're talking about businesses that are built around AI as a core primitive to their existence.

92. 18:32

    Again, from the driver-assisted to the AI automated, uh, driven car. And that shift affords a ton of opportunities, but it is a shift. It's just gonna create a lot of friction.

93. 18:46

    But I mean, just, just to tell you, I... Just with us, we've been around for, you know, 15 or 16 weeks. Small team, six, seven people, and, uh, we built an entire agentic cloud infrastructure from scratch in just a few weeks, and that's unheard of.

94. 19:02

    Like, if you'd asked me that a few years ago, I'd say, "There's no way. There's no way." But with all of these tools and these agents, these agents that we build ourselves for our things, and then, uh, Devin and some others, man, we've made so much progress on this.

95. 19:17

    So it's a huge shift. And especially for founders and tech leaders, the challenge is to, I think, fully embrace the paradigm shift. If you're an experienced founder, this is especially true, but it's especially hard because you've got all this back, uh, years and years and years of experience, and you might need to check some of that experience

96. 19:38

    at the door. Seriously, you might need to rethink, like, you have all this built-up experience that might no longer be valid, or only parts of it are valid, and you have...

97. 19:47

    It takes some critical thinking to analyze that, uh, [chuckles] 'cause otherwise you just get stuck in the old way of doing things. I mean, uh, PwC of all, of all places, they, they had a really good report recently, and one of the comments they said was, "If you're only using AI for a small efficiency gain, then you're falling

98. 20:06

    behind," 'cause there's companies that they're not just using it for a small efficiency gain. Uh, so my admonishment would be, uh, start from first principles. Take this opportunity to step back, reimagine, and refit your company and your culture for this future.

99. 20:22

    You rewire the entire process so that human-to-agent teams can scale that impact exponentially. I mean, that might mean, again, this is friction, this is pain. That might mean redesigning your org chart.

100. 20:37

     It might mean redefining roles and rethinking what skills you hire for. It's a lot of change, but it also can be your unfair advantage if you do it. So with that, I'll just leave you with one question: Is your company using AI, or is it ready to be built around AI?

101. 20:55

     Thanks for listening.