GitOps Grows From a Few Clusters to Entire Fleets
Enterprises no longer run five or 10 Kubernetes clusters. Many manage hundreds or even thousands. That scale makes fleet management a core part of modern software delivery.
Akuity CEO and Co-Founder Hong Wang examines that shift with Alan Shimel at KubeCon + CloudNativeCon in Salt Lake City.
From Argo to Akuity
Wang is one of the original creators of the Argo project. His startup was acquired by Intuit, where the team built Argo CD. They used it to move Intuit’s workloads from private data centers to Kubernetes on AWS.
The project was open source from day one, and the community grew quickly. Wang and his co-founders later left Intuit to start Akuity. Their goal is to help other enterprises adopt Argo CD at scale. Wang says teams can start with a SaaS version and a free trial.
An End-to-End Delivery Platform
Argo CD is only part of the story. The team also created Kargo, a continuous promotion tool. Together, Argo CD and Kargo form an end-to-end enterprise software delivery platform.
Akuity builds on the open source projects with an agent-based, distributed architecture and enterprise features. Wang notes that Meta shared on the KubeCon keynote stage how it uses Argo CD and Kargo for AI clusters.
Shipping to Hundreds of Targets
At KubeCon, Akuity highlighted Fleet Management for Kargo Enterprise. Deploying to five targets is easy enough to babysit. Deploying to 1,000 requires a different mindset.
Fleet Management adds progressive delivery across the whole fleet. It also adds verification and failure threshold controls. Teams can roll out changes in waves and stop problems before they spread.
Kubernetes as the Home for AI Agents
Wang says nearly every conversation at the show was about AI agents. He sees agents as the next workload type after VMs, containers and microservices.
He believes Kubernetes is the right platform for them. Its multi-cloud, multi-cluster foundation made it the natural choice as AI took off. As agent workloads grow, fleet management will matter even more.
Transcript
Hi, everyone. We're back here live with our KubeCon day three coverage. We've got just a few more hours left.
We're trying to fit in a few more interviews. Some people have been waiting all week. We've just been so backed up.
My next guest is Hong Wang. Wang, like W-A-N-G. Hong Wang.
Hong is the co-founder, founder of AkuitY. Yeah. And CEO.
But also, they've been involved in Argo CD, was co-founders there. We're going to hear all about it. Hong, welcome to Techstrong TV.
Yeah. It's great to have you here, man. Thank you for having me.
My pleasure. Yeah. Give people a little bit of your background, if you don't mind.
Yeah. So my name is Hong Wang. I'm the CEO and co-founder for AkuitY.
So I have been working on the Argo project for 10 years. So I'm one of the original creators of the Argo project. And we actually was a startup.
We got acquired by Intuit, and then we left Intuit, actually created this company called AkuitY. I remember. Yeah.
That's our story. Wasn't there a gal involved in the company when they went to Intuit? I interviewed a woman.
Could be Mukilika. Could be. So, yeah.
Yeah, I think- She was our product manager. Yes. Yeah.
Smart woman. Yeah. And then, Dan Garfield's a friend.
I interviewed Dan recently. Yeah. He was involved in Argo, too, right?
Yeah. He's not involved from the very beginning, but he was later joined us as a contributor in the community from another company. Yeah.
Yes. Excellent. So look, our audience really doesn't need an introduction to Argo.
Yeah. It's the third most popular project in CNCF. Really, the whole GitOps thing is around Argo.
Yeah. But if you don't mind, connect the dots from Argo to AkuitY. Yeah, I mean- How'd you get there to here?
Long story short, so we created Argo in a different company. Then later we got acquired by Intuit. Right.
So we joined Intuit, and Intuit asked us, hey, they want to move everything from the private data center to the Kubernetes on AWS. And we were the platform team for the whole Intuit, so we created the Argo CD to solving the multi-cluster delivery problem. And where we are solving Intuit the problem with the project is open source from the day one.
The community is growing bigger and bigger. And at the peak, we are managing about 300 cluster, about 20,000 applications, and providing the end-to-end developer experience for 4,000 developer at Intuit. So it was a great journey.
QuickBooks, TurboTax, all the multi-billion business was running our platform very successfully. So we took that experience, understanding, and we left Intuit, actually create this company for AkuitY. And we are helping more and more other enterprise adopting Argo CD- With that kind of scale ...
in a large scale. But Argo CD is just one part of our story, because later we created a new project called Cargo. It's a continuous promotion tool.
So with the Cargo and Argo, finally we get this end-to-end enterprise software delivery platform. And I just want to mention that yesterday in the keynote of the KubeCon, Meta was on the stage. They saying they are using Argo, Cargo, managing about 100 of their AI clusters.
Really? Yes. That's fantastic.
It's interesting, I remember when AWS first launched. Mm-hmm. " Right?
Mm-hmm. " This is the same kind of thing. If you want to run your business at a scale like of an Intuit- Right ...
which is on Argo, AkuitY allows you to do that. Right. It's that same sort of pitch.
That's the idea. Yeah. Let me talk a little bit more.
So Argo's still open source, obviously. Yes, it is. But I'm going to imagine the AkuitY software is more like maybe an open core kind of thing where- You can say that, yeah.
So what we are doing is we are doing a lot of innovation on top of the open source project. Basically, we turned the Argo CD into a agent base that distribute architecture, and with more any other enterprise feature. Basically, our solution is very suitable for the people in the enterprise and with ever increase infrastructure to manage.
Right now, we actually have many, many customers, including the CoreWeave, Thinking Machine, Poolside, TypeSafe AI. They are using us to managing their AI infrastructure. Love it.
Now, before we get into announcements and into the show here- Right ... I want to give people the on-ramp. Mm-hmm.
" Right. " How do they get started? What do they need to do?
Yeah, just go to our website. io. You will find all the information, and just click contact us, and then you will getting the full support from us.
We do also support a free trial, so you can just log in and you get any 30 days, can try out our functionality. So that's what they do. Could they just download and set it up themselves, you think?
We do providing the SaaS version. It's actually even easier. Mm-hmm.
So within five, 10 minutes, you should be able to see the end-to-end value of our platform. Really? Yeah.
That's great. And it's AkuitY dot? IO.
IO. Important to remember. All right.
Let's turn to this year's KubeCon. Right. What's news from AkuitY?
Yeah, I think one big thing we're announcing is the fleet management capability from the Cargo Enterprise. I think it's all coming from our mandate from our customer. Because most of our customer is actually the big enterprise, so they are not only have five cluster, 10 cluster.
A lot of them has 50, 100, even 1,000 cluster to be managed. So when you deploy a software to one or five target, it's relatively easy. You can babysit it.
You can even do manually do it. But when you're doing that to 500 or 1,000 target, you need a different mindset. That's why we're providing the fleet management to providing the progressive delivery across all your fleet, and also doing the additional verification, doing the additional failure threshold control on top of that.
Got it. That's fantastic. And that's now?
That's now. Okay. Let me ask also about the show.
How have the conversations been? It's all about AI agent. Everything.
It's all AI agent all the time. Yeah. I think the AI agent is becoming the new breed of the workload you are running and managing.
So we are talking about old days, we have the machines, then we talk about the VMs, we talk container, we talk microservices. I think the next stage is about AI agent. And it's a fleet of them.
Yeah. More than a fleet, it's armies of them, right? Exactly.
Let me ask you a question, because we had a little discussion here earlier on our Techstrong gang. Yeah. Is Kubernetes the right platform to be running all these AI agents on?
I think the market speak for themself. In Tuesday keynote from NVIDIA, they asked the audience, how many percentage of the AI workload is running on Kubernetes? Guess what?
What's the number? 99. Yes.
90. Right, 90%. 90%.
So I am of the opinion that Kubernetes is the platform for AI agents. Not everyone here agrees with me, though. So I just wanted to put that out there.
It's my religion. I get it. Hong, all seriousness, I do think we wouldn't be able to do what we're doing with AI and AI agents right now- Right ...
had we not built it on this existing stack- Right ... that we've been building for 15 years. Right.
And that is the point. I think Kubernetes providing a very solid foundation for the multi-cloud, multi-cluster management. That's why when the AI is coming, surfing up, that's why they naturally pick the Kubernetes as a solution.
Yeah. Yeah. I agree.
Hey, I want to thank you for coming by here. For sure. Thank you.
I appreciate your time. io. IO.
Go check it out, especially if you want to scale, they know how to do it. No doubt. You're watching Techstrong TV.
We'll be back in a minute.