Almost every AI company is boasting about their AI advancements and how they impact the future of work. At CREAO, they're living it today. From marketing to development, the entire company runs on CREAO agents. Around 25 people, mostly recent grads, no middle management, projecting millions in revenue. I wanted to understand how that holds together, so I had a coffee chat with co-founder Clark Gao, who has been building CREAO for almost three years, to find out.
Clark is a former data scientist who spent the last years demoing early AI systems to Fortune 500 companies before the word "agent" meant anything to anyone. He has opinions, and he was willing to share the ones that don't play it safe.
TL;DR: An AI-native company is one where agents handle the coordination, the status tracking, scheduling, and hand-offs, so people focus on judgment and design. CREAO runs this way: a small group of people, mostly recent grads, no middle management, with senior "architects" designing the systems and "operators" running them.
Why build CREAO now?
Jen: Let's start at the beginning. Why does CREAO need to exist, and why now?
Clark: It's been almost three years since we started, and the mission has been the same the whole time. We want everyone to be able to use AI easily, without friction. Before this, my co-founder Kai and I spent a lot of time pitching Fortune 500 companies, doing demos back in 2023 when the models were still shaky. There was no such thing as an "agent" yet. But people were already interested. They kept telling us, "I want to build this AI system, I just don't know how."
Jen: So the company exists because of that gap between "I want this" and "I don't know how"?
Clark: Right. If you look at the history of technology, every wave, cloud, big data, machine learning, large language models, the business world takes time to adapt. And to help them adapt, a set of infrastructure companies show up with the tools, the methods, and the services. That's what we wanted CREAO to be. The infrastructure layer that bridges a new technology and the businesses trying to adopt it. That's still where we are, even though the product and the go-to-market have shifted a few times.
Jen: Shifted how?
Clark: We started closer to consumers, and now we want to reach SMBs and eventually enterprises, without losing the professional and individual use cases. The tension is real. Enterprises worry about cost, integration, security, reliability. Solopreneurs just want to get something done today.
Who is CREAO not the right fit for?
Jen: Who wouldn't CREAO be good for?
Clark: The ‘get-rich-quick’ builders. That’s one group I don't think CREAO is great for, even though a lot of them sign up. The problem is, if someone's whole plan is a quick flip, CREAO isn't the ideal tool, because what CREAO does best is something we call compounding AI. It remembers your context and runs a self-improving loop, so your next build is better than your last.
Jen: So the value shows up over time, and they're not around long enough to see it?
Clark: Exactly. They don't have a structured, organized way to compound anything. They do one thing, then next time they come back and do something completely different. It's a different pattern of use. CREAO still enables them to do things they couldn't do before, which is why it attracts them, but they won't get far with that mindset.
The people who benefit most from CREAO are the ones who treat it like a system they can keep adding to.
What does CREAO do that other AI agents don't?
Jen: How is CREAO different from the other agent tools?
Clark: Compared to the general agents out there, our differentiation is that we deliver a cloud agent and manage its whole life cycle, from creating it to improving it to the point where it does the job better and better. We're good at building what we call the agent harness. That's our strong suit.
There are also vertical SaaS tools that bolt on some AI. The limitation there is range. A marketing AI agent can't help you write better code or produce a video. CREAO does the whole workflow, multi-step, long-running, across domains. We're broader, and we maintain more infrastructure around the agents themselves.
What is an AI-native company like day to day?
Jen: How is CREAO changing the way people actually work?
Clark: On the individual level, it extends my skill set. I can't do design, but right now I do all the design work for our conference materials. I ask CREAO instead of going to our designer, because the designer has other things to do, and I deliver the project end to end myself. Same with data. We integrated our database so anyone can just ask the agent about our daily operations instead of the old way.
Jen: The old way being a product manager going to a data engineer, and three people sitting in a meeting to figure out what happened?
Clark: Exactly that. I was a data scientist, so I know that process well. Now it needs far less coordination.
Jen: What about the org level?
Clark: It changes how we manage people. CREAO is still an early-stage startup, but not a tiny one. We've raised $30 million and we have a couple hundred thousand users. And we do that with around 25 people who aren't even sitting together in an office. Everyone works at their own pace, across time zones. A lot of people join and say, "This is an odd company, I don't even know who my manager is." But they still deliver, because the agents actually enable people to do the work.
Jen: Where did this structure come from? It's not like there's a playbook for AI-native companies yet.
Clark: I thought about this recently. What I'm doing in go-to-market is close to a consulting or law firm partnership model. I see each channel owner as a partner. They don't just work on a project, they bring in the project, build their own small team, and own the result. We aggregate what each of them delivers. Peter, one of our leaders, wrote about the shift a few months ago. His theory is that going forward you mainly need two kinds of people. Architects, the senior people who design the systems and agents, and operators, the people using those agents to deliver results. The layer that thins out is middle management, the roadmap-and-status-update layer.
Jen: Is that why you've hired so many junior employees?
Clark: Yes. Several of our go-to-market people graduated last year. When the agents handle coordination, a sharp junior person with the right tools can own a whole channel. It's ambiguous at the start, and that's uncomfortable for some people, but this is the very beginning. The guidelines and best practices will come as we go.
How should you think about AI and your career?
Jen: How do you address the fear people have about AI?
Clark: I got the same question last week on a panel. My honest answer is that I'm an optimist, partly because I understand the technology and have worked on it for years. The first thing anyone should do is learn how AI actually works. When I interview fresh graduates, my first question is which AI tool they used at school, and a lot of them say their professor didn't allow any AI during exams. If you don't know how to use AI, it's going to be hard to keep up here.
Jen: So the move is to actually learn it instead of avoiding it.
Clark: Right. Get to know it. Once you understand it's a probabilistic model, that it's statistics underneath, a lot of the fear settles down. And I genuinely believe AI creates more demand, which over time creates more kinds of work. We're in a transitional phase, but new roles keep appearing that nobody had imagined.
Jen: What's the skill that actually matters going forward as more and more companies adopt AI?
Clark: Critical thinking. For decades everyone chased engineering, software, or data science degrees. Now, when we evaluate talent, we look at how someone reasons. We look at people who studied social science, history, philosophy, or the arts. If you can write well, read fast, and hold your own in a debate, including a debate with an AI, and still form your own ideas without being pushed around by what the model tells you, that's the valuable skill set now.
A lot of the people building frontier AI came from psychology, philosophy, neuroscience. That range is what lets them see where the technology could go.
The businesses getting the most out of AI have one thing in common: they're building a system that compounds, whether that's Charis turning a few prompts into a video pipeline, or Clark running a company where a 25-person team operates like a much larger one because the coordination layer is handled.
This is the latest in Behind the Agents, a series where I sit down with CREAO team members to understand what an AI-native company looks like day to day.
If you want to see where this goes for a solo operator, Clark's advice maps almost directly onto how to build a business that runs on AI. Start with one workflow you'd want to keep. Build the agent. Let it compound.
Related reading
Frequently asked questions
What is an AI-native company?
It's a company built so that AI agents handle the coordination and repeatable execution, and people focus on design and judgment. At CREAO that shows up as around 25 people running many channels at once, with agents doing the status-tracking and data work a middle layer used to do. You can see the operator's view of that in managing a team of agents.
What is an AI-native platform?
A platform where the agent, its memory, and its improvement loop are the product, rather than AI features bolted onto existing software. CREAO's version manages the full agent life cycle, so an agent gets better each time you run it and can carry a multi-step workflow across domains instead of being locked to one task. Clark calls the reliability layer around this the agent harness.
Should I learn AI even if I don't work in tech?
Yes, and Clark's advice is to start by understanding how it works rather than avoiding it. Once you see that a model is statistics under the hood, it feels less like magic and more like a tool you can direct. The skills he values most now are writing, fast reading, and clear reasoning.
Is CREAO a good fit for solopreneurs and small teams?
It's a strong fit if you want to build something that compounds over time rather than a one-off. The value comes from CREAO remembering your context and improving with each run, which rewards people who keep adding to a system. Here's the walkthrough for building a business that runs on AI.
How do you build an AI-native company?
Start with one workflow instead of a transformation plan. Clark's version at CREAO: agents take over the coordination layer first, the status tracking, scheduling, and hand-offs, then senior people design the systems (architects) while everyone else runs them (operators). The structure follows from the agents, and the playbook gets written as you go. The walkthrough for a solo operator is in how to build a business that runs on AI.