The Future of Trusted AIOps

Hey, everyone. " My next guest, and I'm going to do my best to get this name right, is Xavier, that wasn't a hard name, but Lecrochaire. Lecrochaire.

As close as my New York accent's going to get to it, I think. Xavier Lecrochaire, or Xav as his friends call him, is senior director, Ansible Automation Platform Delivery for Red Hat. " It's great to have you on here, man.

Thanks for having me on. Really appreciate the invitation here. All right.

So, we get from the name and your accent, you sound French. Yeah, from France? Yes.

Tell us how you wound up here today. What's your journey been like? Yeah, so thanks for the question.

So yes, I'm French, but I've been in the United States for over 30 years now. I'm based out of Chicago. My background actually is all around configuration management and automation.

So for the last 30 years, I've been working in the software industry, mostly as a product manager, starting in the consulting business side, in the field, learning from customers. And really early on decided that really like to understand and solve customer problems through products and software products specifically. So across the years, I've got to with multiple companies, all in the domain of automation ultimately, because that's where we are today.

And now where I'm at, I'm working for Red Hat. I'm in charge of product management for a product called the Ansible Automation Platform. Right?

And I'm in charge of, basically my job consists of learning from customers, what are their needs in the domain, and how I can shape the product roadmap and deliver the product that meets those needs. So I'm really passionate about ultimately product management and the domain of automation. So that's kind of highlight of my background here.

Excellent. Excellent. I mentioned in the introduction, you're the senior director for Ansible Automation Platform Delivery.

com. We don't need any introduction to Ansible. com, well, March 2014 was the first publications there, but Ansible was then in 2014, and it's here now.

And when you look back then, we used to say Chef, Puppet, and Ansible. They're both still out there, right? But Chef and Puppet are not necessarily, I think, what they were 12 years ago or whatever.

Ansible for a lot of ways is, right? I think if you look over the arc of time, it's held its position. But of course, everything changes.

Everything changes with the technology around it. So when we look at Ansible Automation Platform Delivery, what are we really talking about, Xav? Yeah, that's a great question and great point.

So first of all, you mentioned Ansible is still here, right? That's the first point you made. And I think part of the reason why it's here is because the simplicity of the automation language that Ansible delivers is front and center to the Ansible automation platform.

Not only that, we talk about human readable, and we'll talk a little bit about AI probably later, but it's very important to be able to understand what this automation is going to do. Right? And the second point I want to make around why Ansible is still here today is because really Ansible is about automating ITOps across all the domain of IT.

So it's not just patching, it's about being able to thrive and being able to solve networking problems with automation. How do I connect all those domains together? Ansible help at addressing those and unifying the language of automation.

But that's the language that the foundation, right? For me, the Ansible automation platform is a lot more than just, I'm automating this execution. Our customers need to build that automation.

We need to help them build all those playbooks, all those roles and collections to interact with the third-party systems, right? So that's a critical part of what we deliver. But also our customers are really focused on solving problem at scale, enterprise problem solving at scale.

Right? So it's like is the governance, the platform itself delivers the governance that our enterprise customer need, right, in order to manage effectively their automation, the ITOps automation at scale in their environment. I love it.

That was excellent. Thank you very much. Can't go too far into discussions around software development without mentioning AI today.

It's just that's the way it is. Yeah. So like everything else, Ansible has had to, or not Ansible itself, it's not a person, but you and the Red Hat team have continued to develop Ansible to take advantage of the latest AI, AIOps to action, as I think you call it.

But all of it, it's got to say state-of-the-art. It's got to stay state-of-the-art. That's the important part.

And we talk about the future of trusted IT operations. How are you doing that, Xav? And where do we see it in real life, you know what I mean, in plain sight?

Yeah, so that's a great question, and I think there's different areas of AI that we can discuss in this particular point. So I'll start where we went with the Ansible automation platform is historically our first foray with generative AI, and I'll start there before I go to agentic AI, right? We went into a project three years ago, that was prior to ChatGPT, together with IBM Research, looking at, hey, how can we leverage AI to help people create that automation, write those playbooks, right?

So think about that. So we went into a journey with them and introduced that as part of our platform. That was the first step.

But as AI started growing and growing, what we're seeing is we're seeing opportunity for AI, generative and agentic AI, to really help in the automation domain and the journey of automation. What I think it is, is like AI, people still do not necessarily trust, and it's going to take a long time before you get to that autonomous AI kind of objectives that people are talking about. I don't think AI removes operations by any means, right?

And people still need to apply those operation and deliver them with governance, right? That governance that goes with automating mission-critical IT systems at scales. But what it is, is where we think At least where I think AI can help is really to help analyze that monumental amount of data that now is at the fingertip of people, to provide set of recommendations on, hey, if this is the type of data, if this is the signals that I'm receiving, here's the type of automation, here's the automation that I need to execute.

So I think AI can help bridge that kind of decision, if you want, the time that it takes to react to a set of signals to execute that deterministic automation that I trust. I've invested in building a deterministic automation, but how can I bridge that from the signals that are happening, to this is the automation that I need to execute right now to solve the problem, and sometime proactively even solve that problem that I may face in the not-so-distant future. Excellent.

Look, I speak to a lot of people. I speak to vendors, I speak to practitioners. When I speak to IT leaders, and I don't want to paint with a broad brush because everybody's different, we're all going along this journey.

We're all going along it at our own pace. Yeah. Specifically for leaders out here, how can they leverage this?

How could they adapt? How could they stay on top of the game? So I think it's really about, for me, how can they stay ahead?

It's really about being able to introduce AI in their processes, in areas that really helps them right now. And as I said, I think it's important, so I talk about historically our first foray with generative AI was helping people, if you want, create automation faster. It didn't remove how people were actually validating that automation was in a line and was governing what they were doing.

It was the creation to help that creation. Where we're seeing now the shift is about being able to leverage AI, to analyze the data and provide a recommendation, but not remove necessarily the human in the loop. That is, sure, I understand what needs to happen, go ahead and automate these particular processes for me or this particular set of actions that I need to take.

We're not going to see that. To me, I think this is very important to understand that is, AI doesn't remove operations and doesn't necessarily remove human in the middle or human in the loop as part of what needs to happen. There's too many risks associated with that.

We see people blindly accepting recommendation from AI to do certain things that suddenly has cascading effect. Also, in the domain that we are is in some cases, I'm going to use basic example. Patching a system may have wide effect that people don't necessarily understand if they haven't done the analysis, and it might affect an entire set of other systems.

So that part, I think is still a problem that AI hasn't necessarily solved fully, and where people, you still need to have the critical thinking that goes into those recommendation. But to me, the recommendation from people is consider where AI at first to introduce AI, where it will help you solve immediate problem and where it might help people really understand a broader set of issues or a broader set of opportunities that exist there. It cannot remove, it has to be able to interact with your governance policies, like from an enterprise point of view, people have, let's stick with the patching example.

I've got corporate, windows of upgrade that I want to maintain cannot go blindly. So all these things together is where AI can help analyze the broader set of data that exists out there. But ultimately, you really want to execute the automation that you have trusted for years, and that as you grow into that trust, you can bring more elements into the view of AI to help with the plethora of data that exists out there.

Because it might be that there's multiple factor that will help you more quickly address some problems that you might have in your environment. Sure. I think it's not simple.

Also, AI has a big cost as we start talking about tokenization. And I think so it's a balance of looking at those things. And really for me, I want people to feel like they can onboard and can get the value of AI without having to completely rip apart all the investment that they've made over the years.

And that's a critical part, that I don't feel of I have automated, and you mentioned before, my favorite automation, whether it's Ansible, Chef or Puppet. I've got these. I may not have all the skills there, but I still trust these, and this is what I need to use in order to execute that automation.

AI might help me make the right decision in terms of this is the one that I need to use right now. Sure. Look, we're recording this on a Monday.

We recorded this Monday. Last week, the wires were full with the story of this OpenAI agents breaking out, breaking containment. You can't blame the agent.

They do what they're programmed to do, looking for information on how to solve a problem, and they went to Hugging Face because that's where they felt they get that information. But it freaked people out a little bit. This thing went off the reservation, so to speak, and broke in using yet another zero day that it used, broke into Hugging Face to get it.

I don't know as a response, and also last week, we saw two of the Chinese open weight models. They're not open source. Open weight models released their newest versions, which rival Claude and OpenAI.

Then we saw Jensen Huang put his name to a letter, I think IBM is on it too, with Red Hat, about that AI needs to be open. We need to have open AI. And then today, news broke about this open secure AI association.

I don't know if it's going to be like a Linux Foundation kind of foundation or what have you, but an alliance around secure open AI. Plainly, clearly, the industry is calling out saying, "Hey, we know we got to do this. We got to be open.

" This is a story Red Hat's been talking about for a long time. Before there was AI. Open and secure.

This is up to the minute stuff, Xav. I don't know if you've had a chance to internalize it even, but how does that play in here? Yeah, no.

So I think you're making a good point. Yeah, I was reading about the OpenAI and how it escaped the sandbox to go and do a bunch of intrusion and things like that. And I think this speaks to me to governance, and how people need to think right now, especially in the domain of ITOps.

On how they go and they address problem in their environments that are mission critical. I think it's fundamental for the AI world to reflect on this. I know that there has the Mario long, how do you call that?

Essay about this. About we need to be really careful about where this can take things, because obviously the agents, everything is becoming more independent, more effective, but also being able to build a governance that goes into how an AI needs to start dealing with a set of problem is fundamental. So I think we need to really think about it.

The problem is we also need to anticipate the fact that not everything that will be AI will be ethical. And that's a part of the challenge. But being able to work together with our ecosystem, with our partners around how AI can be done and leveraged ethically speaking, is very, very important.

We talked at IBM Red Hat about Project Lightwell, which came from the Mythos moment of, well, now suddenly the AI can understand and highlight a bunch of security issues in the software. And I think that a lot more faster, sorry, and in a lot more massive way, and we've seen it ourselves. But where I think Ansible automation platform can help, for instance, and AIOps in general techniques, is really understanding that we can patch those systems faster, as a result of being able to not only as a result of leveraging AI, but leveraging the entire automation platform to patch those systems.

I think that's very important. But I think the conversation around the ethical use of AI, the data that needs to be used, the open source aspect of things is absolutely critical. And this is why, many years ago, I came into the world of open source.

I was in closed source software for many years, and I've been now at Red Hat for over almost 20 years now. And I think that's the power of open source software is being able to level set across the data across all our customers, and being able to leverage that software to understand where you have risk and where you don't have risk. So, for me, I think AI will push the envelope, but we still need to understand that AI doesn't remove operations at this point.

We still need all those operations. Absolutely. Xavier, we're past our 15 minutes to tell you the truth, but I wanted to end.

People who want to get more information about Ansible automation platform, what's their best website or best place to go? com website, Alan, and be able to look for Ansible automation platform. And Ansible automation platform is one of the four main platform for Red Hat with OpenShift, RHEL obviously, and now OpenShift AI.

So, and together the Ansible automation platform and OpenShift AI are doing a lot of work around those domains that we're talking about to help with the customer adoption of AI in a safe and secure way. Love it. Xavier, thank you for coming here on Techstrong TV.

It's been a pleasure. Thanks, John. Hope to see you back soon.

Keep up the great work. All right. Sounds good.

Thank you, Alan. Nice meeting you. All right.

Bye. We're going to take a break. We'll be back in a minute.