Why Building AI-Native is the Biggest Opportunity of this Decade

The Signal

If you were forwarded this newsletter, join 4,095 weekly readers—some of the smartest GTM founders and operators—by subscribing here:

---

Hey y’all, and Happy Friday! 🤘

There’s a thesis that keeps bouncing around in my head, and I can’t shake it.

I haven’t fully formed my thinking around it beyond “I think this is very important” (I don’t know exactly how this plays out). But, I wanted to write out my thoughts on the subject - or said better: by publishing something on this topic publicly, I am forcing myself to clarify my thinking around it.

The concept is building AI-native, both as a company and as a worker. I think it’s the biggest opportunity of the decade. Heavy, I know.

Okay, let’s get right into it.

---

Why Building AI-Native is the Biggest Opportunity of this Decade

The biggest breakthrough in AI is building a company in an AI-native way. I also think you can become an "AI-native worker.” Today's post explains this concept, and why it’s such a unique opportunity.

There are a lot of little use cases with AI. Some are good. Some are great. Some don't work (even though people online say they do). Some of these are just basic prompts. Others are very robust AI workflows or even full-on AI agents.

I have a note on my desk as a reminder to "build AI native.” Proof below (not AI-generated):

AI-Native Companies

This is the best definition I’ve heard of being “AI-native”:

As models improve, my customers' experience improves.

If I were building a software company today, I would build it in a way that as new models are released (for instance, Sonnet 4.0 came out this month, ChatGPT's O3 recently came out, and Google Gemini released their latest version model at I/O last week) my product gets better automatically.

Said another way, if I'm building on top of these models and architecting my product to leverage these models, I literally have the best engineers in the world improving my product every single week. This means that my customers' experience (automagically) improves as the underlying models (inevitably) improve. This is nuts.

I keep re-reading a message that a founder texted me earlier this month. He is someone I deeply respect, is incredibly smart, and is currently building an AI-native product that’s in the top .01% of companies. He said:

It’s unbelievably easier today setting up a company and start making profit than it was a year or two ago.

In 2-3 years software is going to be so commoditised that it’s going be hard to build something and sell it with a competitive advantage.

So I’d bet the best timing is now.

Think of building something that’s riding the trend - i.e. that with every model update your product or services gets better.

He’s describing building a company that is “AI-native.”

If this doesn’t get you amped up, I don’t know what will.

AI-Native Workers

The other way that I'm thinking about this is on a personal level. Becoming an “AI-native worker.”

To illustrate the point, here are some real examples in my day-to-day workflow, that I’m using AI:

• Generating an SoW

• Editing a blog post

• Building a GTM workflow

• Using a prompt to ideate signals for a GTM team to use

• Action items+next steps+owners (created from a call transcript) for my consulting clients

What’s amazing is, without changing the prompt at all, every time I use that exact same AI workflow, the output—a.k.a. my work—will be better because the underlying model has improved since the last time I used it. This is a way that I have never worked before.

Basically, this means that I'm moving from a world of using a template, let's say for a statement of work or a proposal (replacing company name and manually replacing the scope and price and start date) to a world where I’m using prompts (and the entire proposal is dynamically generated and the output of it/quality of it is better every time because the underlying models are improving). Utter magic.

The Trap to Avoid

The reason 99% of people reading this won’t actually work this way, is because 9 out of 10 times doing a task "the old way" is way faster than doing it using AI. It’s why I have that silly note on my desk — to force myself to work in this new way.

I think anyone who has seriously played with AI outside of just having it generate a grocery list or some other basic use case realizes that to truly get these models to work consistently, you have to do a lot of work: whether that’s tweaking the prompt, connecting multiple tools, or even just figuring out which tool or model is best for a certain job.

I assume if we fast forward two years, all of these sorts of decisions will be abstracted away from us, but today you have to do a lot of extra work to see the outsized returns with using AI.

I was inspired by Kyle Norton who recently shared a memo around the urgency of AI adoption. And I wanted to do the same (also, I’m very much so speaking to myself here with this post). Note: I don't have anything to sell you. I promise I’m not going to launch an AI course (I will never sell a course). I don't run a SaaS company or an AI SDR company that leverages AI. I'm just trying to learn these things. And, I see The Signal as a sort of “diary” of my learnings that I share publicly.

Over 5,000 people will read this post. My hope is that even just 1% (or, 50 people) take action and jump into building in an AI-native way. Either an AI-native company or becoming more of an AI-native worker.

AI is Similar to Managing an Intern

The best analogy I’ve found for working with AI is like working with an intern. If you've ever worked with an intern (or managed anyone, really) you probably have experienced the realization that it’s often easier to just do a task yourself instead of telling them how to do a task or correcting their work.

But, once you get over the hump and document all of the steps of a task, the intricacies, the nuance, and the goal or the outcome — what you realize is even if you technically could do the task better, when you offload that task to that junior level person, you gain incredible leverage.

In my experience, this is very similar to how working with AI works today. The difference is, I would argue, that a lot of the output—and I don't mean to offend anyone—but a lot of the output of the AI is actually higher quality than what an intern or a junior level employee might produce. So putting in this extra work to get over this hump not only creates leverage on your time, but it actually, I've seen, gives you the ability to generate things that weren't possible before.

Closing thoughts

So whether you're building a company or just trying to use AI more in your day-to-day workflows, I encourage you to keep pushing. PS - reply to this email and share any practical ways you're using AI today.

I'll leave you with one little, practical thing. I use a standalone app on desktop and mobile called Superwhisper that allows me to dictate very accurately. One of the other things that I do with Superwhisper is I'll just talk to it for several minutes. Maybe it's specs for something I'm trying to vibe code. Maybe it's a consulting proposal. Maybe it's a follow-up email after an in-person conversation. Or ideation around go-to-market experiments. Then, I take the big block of text and drop it into Claude or OpenAI and have it generate an output based on my goal. I found that this is super useful and I do it once or twice a day now. In fact, as an experiment, for this article, I dictated almost the whole thing, and then I just copied it into Claude (Sonnet 4.0) to clean it up and then dropped that into the Substack editor. I only changed a few sentences. So if this blog post sounds very conversational, it's because I said all of this out loud. I didn't write it on a keyboard by typing. I think I can still say this post was not AI-generated 😅 (definitions are getting fuzzier!).

I’m excited to see this wild new world—and how we work in it—unfold right in front of our eyes.

---

As always, thank you for your attention and trust.

Have a great weekend, and I’ll see you next week,
Brendan 🫡