AI-Driven Infrastructure: Ido Neeman on the Future of Cloud Management

Hey, everyone. We're back here, live on the, uh, floor of Text, of Text Drunk. We're text drunk, but we're back here live on the floor of CubeCon and you're watching Text Drunk tv.

It's been a long two days. Uh, let me introduce you to my next guest. His name is Ido.

Neiman Ido is with Firefly. And if you haven't heard of Firefly, don't worry you're going to. But first, IDO, welcome to Techstrong tv.

It's great to have you here. Thank You for having me. So Let's talk about you first.

Alright. Give us, give us the Ido Neiman story. So, uh, I feel at home on the tech strong floor.

Oh, within the bigger, uh, within, um, it's This little gray area right here. Yeah, Exactly. Um, and, uh, I feel at home because, uh, generally I'm a nerd.

Uh, grew up around computers. I served 12 years in the Israeli cyber intelligence 8,200 Or, uh, something little bit better. 81.

Oh, 81, okay. Yeah. Yeah.

Uh, the last, the, the last digits you get the better it is. Okay. Uh, so, uh, then I led technology for prominent Israeli hedge fund.

Okay. Um, then I had a startup in the serverless field, Uhhuh. And for the past four plus years, I'm building fireflies together with my, uh, partner and great team.

Fantastic. Yeah. And, you know, look for those of us in the industry, not for nothing, but that's a pretty typical Israeli entrepreneur story, right?

Yeah. We, I mean, a lot of them for people not in tech, they're like, wow. But, you know, but this is of course why they, you know, Israel's a startup country and, and these kinds of things.

It's, it's, it's what you, you do there. Um, you know, I've been interview, I've started four or five companies myself over the years venture, and I've interviewed hundreds of co-founders and, you know, entrepreneurs, no one starts a company lightly. Right?

Everyone, you put your, your, your, your guts into it, right? And, and then your passion. Talk to us about what was your passion in doing Firefly?

So back in my Army days, we, we obviously did some very complex technology building, but the scale, you know, it's a nationwide scale, right? We did some very large cyber activities, let's put it this way. So always fascinated by doing large things at scale with efficiency.

And back then we were looking at the outside internet and seeing all the tools and wealth of, uh, of possibilities. And because we are, you know, air gapped, we need to build everything ourself to be military grade, we can only build those. So I remember seeing stuff like infrastructure as code coming to life, wanting it, but need to build our own scripts and internal tools to, to, to do it.

So, you know, when I got out after a good run, uh, I started coming to shows like this. Um, and this, this got me pumped about the possibilities, but also to start thinking about how we harness all of this, because it is quite a lot. You said the long two days, and we still have at least one day to go.

Yes. So many different conversations. Vendors, technologies, 500 talks outside.

Yes. So imagine that you, you come to CubeCon and say, Hey, you know what? I'm gonna be cutting edge.

I'm gonna learn everything. You just can't, you can't Anymore. Right?

There was a time when you could, right? Maybe the first one in San Francisco, 2018. Yeah.

Yeah. Not even 20 by 2018, that was already, I think Seattle, it was big 20 16, 20 17 maybe. But yeah, now it's just, it's 200 projects in CNCF alone.

Yeah. Lot to wrap around. So when, when, when you say passion is about, Hey, let's build the best technology, let's make it scalable and accessible and simple, but also let's allow the end users to actually adopt what's happening here.

Because everyone wants to create the new shiny CNCF incubation, uh, project. Everyone wants to be, Hey, I got the next 10,000 stars and everyone loves me and I'm great. Right?

Everyone. But at the end, we want people to use our technology, right? We want the end customers, of the end customers to gain value out of it.

So at some point I said, Hey, I think this is both interesting, hard, and might, uh, might be useful to the world. I love it. Yeah.

Now, if I, if I had to ask you define Firefly's mission. You talked a little bit about your passion and how it relates to that, but crystallize it, you can look, this, this camera's on you tell, give them Firefly's mission. Its reason for being Our mission is to make cloud simple, managed and controlled.

So, uh, Firefly is an agent cloud automation platform. Basically, we help you understand everything that runs in your cloud. Because as mentioned, 200 projects, 500 vendors, so many practitioners, how do you even know what you have in the cloud?

We just saw it, uh, three weeks ago. We're now talking in November. But, uh, in late October of 2025, we saw two major, major, major outages, um, by AWS and then by Azure.

Azure, right? So now we know that everything runs on the cloud. You can say AI is the most important thing.

I don't disagree, but I say AI runs on the cloud, is why you see all those large AI companies sign deals with cloud. Mm-hmm. So cloud is the most important mission critical thing we have in our lives.

For, for us computer nerds. You know, we couldn't get DoorDash to deliver a lunch so we know something bad happened, right? Uhhuh, um, so, so important.

But it's so complex, it's so expensive. It's, it changes so fast. How do we control and automate it?

We don't, we don't have one place, one system of record to see our cloud. So this is cloudent cloud automation platform. We connect to everything in your cloud seamlessly.

Within minutes, we, we see everything from Oracle to A-W-S-G-C-P, Kubernetes, your Datadog, your MongoDB, everything that's clouded. We see it in one place. Then we sweep in and we tell you over the system of code, what's properly controlled and qualified.

What's, what's governed, what's managed, what's looking like you want it to look like versus unmanaged or misconfigured. Once we establish this, we bring you a bunch of automations. It can be full agents, it can be ai, AI driven automations.

It can be quote unquote old legacy, uh, algorithm based automations to solve your problems from deploying to the cloud, to govern it, automating it, and up to recover from an incident like we saw three weeks ago. Excellent. Good.

That was good. ai. You, You, you can't, you can't do it, uh, any other way today.

com. Right? com, though we do have AI and IT ones too.

There you go. You see? Yeah.

Well, we like to spread it around, but, um, let's talk. So now let's fast forward, right? We understand the problem that you're solving a little of the history and passion as the company exists today, right?

And we live, I mean, yes, cloud's very important. Agentic AI is probably be very important. It's very important.

Now will be more important next year and the year after. Where is the company? Like when you're talking to people here on the floor today, what are the conversations you're having?

What are we talking about? First, we're talking about what are the best parties post the conference, right? That's, It's the most important.

That's most important thing. Yes, Of course. The second thing, and if you want to take the AI out, one thing that, that, that, uh, we've noticed and, and we actually learned about it from our customers that we found it, found it, uh, ourselves, is that today, application software is being written by ai.

Mm-hmm. Everyone uses ER or, or copilot or cloud code or, or doesn't matter what, right? But when we talk about cloud infrastructure, it's still being run manually with click ops or line by line terraform open Tofu.

Um, at best it doesn't make sense. How come all my backend engineers and front engineers are writing to Carol, say, Hey, help me do this and that. I'm now writing application and, and the agents do it, but the infrastructure that powers it all is still old way.

It doesn't make sense. So people are looking for ways to integrate AI into infrastructure management. We see it through our MCP server.

So MCP servers allows you to use your cursor, your code code, you get them copilot in tandem with Firefly. So you just talk to your AA system, Hey, I wanna now deploy this application into AWS use is one. What do we need to change?

Goes to Firefly, see? And it solve it. So this is one thing moving from, um, manual scripted cloud ops into AI and agent cloud ops, just like software engineering moved it.

This is the first thing. The second thing, um, as I see it, is how do you incorporate agents into your day-to-day workflows? Because everyone wants it.

Right? After every single board meeting, you'll see the, the CEO coming to the CTO, which come to the vp, which come to the developer. We need more ai.

Now, if you don't adopt ai, you're fired. If you adopt ai, you make mistakes, you're fired. Yeah.

I don't like, so either Way, you're fired. Probabilities are bad, right? Um, so, but what happened?

So if you remember two months ago, a vibe coding solution created an application, deployed it to the cloud, changed it, and then it, it deleted a production data. Lied about it. Yeah.

Yeah. I, I, I, I'm here to talk about the good stuff. Okay.

Okay, go ahead. I live, I live here to talk about the, the later. Yeah.

So deleted it. Now everyone say, Hey, we can't use the agents in vibe coding in production. It's nice for a, you know, your hobby.

I want my, my grandma's recipe application, right? Cool. But I want let it Not mission Critical, right?

My trading of a solution. So what do we need? Because at some point we will need agents to do it because we don't have enough platform engineers.

They DevOps engineers out there SREs to do it. We will have it. What are we lacking?

We lacking context. I need the agents or the AI driven automation to know that this database is being used by other teams made, deployed on other lines of business, other continents. How do I know it?

If I have this system of record for cloud, this is the first thing I know. And the second one is guardrails. So as we see it, when we talk to the market, we see AI is being treated like a very good protege or, or an intern.

You wanted to do stuff, but only to some extent. So you say, Hey, AI deployed to, to dev account, maybe to staging, but you don't deploy to production. So you need a guardrail that says, Hey, agents can deploy to staging, but when it comes to production, you must have a human in the loop.

So it opens a pull request for you to do a code review for it. And so all of these types of, of limitations. So to, to wrap it up, we need to adopt AI in our infrastructure management and operations.

And not just for application, because usually infrastructure move faster than application development. And the second one is have the context and the guardrails to really adopt agents into platform engineering. I love it.

Um, so, you know, the, the, so that's what should happen, right? The world's full of what should happen. Not, you know, but that's not necessarily what happens.

How do, how now and Firefly's mission is to make sure that happens or, or to help make that happen? Enable, Yeah, enable it. Do you see, and one of the things I see, look, I I've been in technology a long time, right?

I got the dot coms. But, uh, there's a gap between what should happen and when it happens. How big a how big a gap is that, how this vision you have of, of how this should work?

When are we going to actually see that? So, you know, uh, many people ask me why we have, uh, agen, angen cloud automation. So the first thing is that they don't let us in into coupon floor, or sorry, text floor without agent in your name, right?

If you don't got ai, you don't have Yeah, Exactly. But because everyone is talking about it, everyone wants it. But we're still in the gearing up stage, right?

We, What I talked about the context in the MCP, this is reality. We have many customers. I'm not talking about the handful.

We have many customers who are today consuming Firefly with an MCP first mindset, really. So no ui, no A-P-I-M-C-P first. And I'm not talking only about the cool kids from Silicon Valley running the AI shops.

I'm talking about some very large enterprises, really legacy enterprises, fortune five hundreds. And I love seeing it because it, it means that it's now, you know, it's mainstream. Yeah.

We, we are at the height of it. And it's, it's, it's great for the actual agents to deploy. It will take some more time because not enough context, not enough guardrails.

And we, we can still, And not enough trust. Exactly. Exactly.

I I, I think that's it too. But that's interesting that you have MCP first. 'cause you think about it, what MCP came out about a year ago now only Makes sense.

Yeah. December, I think last year. And now we think about it, we talk about it like it's been here forever, but it's not, it's not even a year old.

But here you are selling MCP first. I think that, you know, you don't need to tell you how successful those ai, big AI companies, but even the, the, the, the companies like Cursor and Windsurf. Yeah.

They're so popular. And every developer software engineer today ask themselves, how can I get better? How can it get stronger?

And if you don't do it, it's not that you're getting better. It's like, um, inflation, your power goes down, right? Yeah.

So everyone, even just to keep up, you, It's just table stakes at this Point. And, and, and now the platform teams, the SRE teams, they say, Hey, we are now behind because they have it writing code. But what about our infrastructure?

What's the difference code as a new code deployed for application is more, most often new. And it's, it's similar. Every company's, every organization's infrastructure is a snowflake.

And it comes, if you're not a five person startup, you come with a very big baggage that you need to carry with you and understand and tap into. So this is why it was harder for them to do it. So, so if you allow them to run as fast as the application teams, they just embrace it and they love it.

And they come back to CIO, Hey, you told us to adopt ai. We're adopting it. See, how cool is it?

Hey, I'm talking to the MCP server. Show me all drifts in my account. Firefly fixes it.

Hey, here's a misconfiguration. Let's use AI to, to solve it. So Firefly finds a problem.

The MCP creates the solution, then we bake it into infrastructure as code, we deploy it. People just love it. It's, you know, the future is here.

We, we, living in a place where everything can, can, can be quick and effective. It Is, it is an interesting time to be alive. Yeah.

I will say that. Boomer or not Ido. Thank you so much, sir.

Thank you. Eddo Neman Firefly here on, uh, text Drunk TV on the text Drunk TV floor as far as the eye could see. We're gonna take a break.

We're back with more. Stay tuned.