AWS Just Released an AI DevOps Agent - Should You Be Worried?

AWS just launched their AI DevOps agent AWS just launched their AI DevOps agent and I can already see the videos coming and I can already see the videos coming and I can already see the videos coming that DevOps are now replaced by AI. It's that DevOps are now replaced by AI. It's that DevOps are now replaced by AI. It's not worth learning anymore. You can not worth learning anymore. You can not worth learning anymore. You can forget about DevOps. It's all AI now. forget about DevOps. It's all AI now. forget about DevOps. It's all AI now. But when you read the fine print and But when you read the fine print and But when you read the fine print and when you look deeply into what these when you look deeply into what these when you look deeply into what these things actually do, a very different things actually do, a very different things actually do, a very different story comes about. I also just came back story comes about. I also just came back story comes about. I also just came back from KubeCon, one of the biggest from KubeCon, one of the biggest from KubeCon, one of the biggest Kubernetes conferences in the world and Kubernetes conferences in the world and Kubernetes conferences in the world and at KubeCon there were multiple companies at KubeCon there were multiple companies at KubeCon there were multiple companies advertising AI SRE and agentic advertising AI SRE and agentic advertising AI SRE and agentic observability or automated incident observability or automated incident observability or automated incident resolution. And I really took time to resolution. And I really took time to resolution. And I really took time to watch these demos and what these watch these demos and what these watch these demos and what these companies were doing so that I can give companies were doing so that I can give companies were doing so that I can give you a my take on this. Is AI going to you a my take on this. Is AI going to you a my take on this. Is AI going to replace DevOps? Now, I'm not here to replace DevOps? Now, I'm not here to replace DevOps? Now, I'm not here to fear monger, but neither am I here to fear monger, but neither am I here to fear monger, but neither am I here to paint a rosy picture. I want this to be paint a rosy picture. I want this to be paint a rosy picture. I want this to be an authentic discussion based on facts an authentic discussion based on facts an authentic discussion based on facts and from what I'm seeing in the and from what I'm seeing in the and from what I'm seeing in the industry. So, the question is, is AI industry. So, the question is, is AI industry. So, the question is, is AI going to replace DevOps engineers or not going to replace DevOps engineers or not going to replace DevOps engineers or not and where are we at currently? So, let and where are we at currently? So, let and where are we at currently? So, let me tell you what the AWS DevOps agent me tell you what the AWS DevOps agent me tell you what the AWS DevOps agent actually does and how it this is going actually does and how it this is going actually does and how it this is going to affect us as DevOps engineers. But to affect us as DevOps engineers. But to affect us as DevOps engineers. But before we continue, my team shared this before we continue, my team shared this before we continue, my team shared this very important statistic with me, which very important statistic with me, which very important statistic with me, which is that 78% of those who are watching is that 78% of those who are watching is that 78% of those who are watching this video right now are not subscribed.

this video right now are not subscribed. this video right now are not subscribed. Some of you have been watching my videos Some of you have been watching my videos Some of you have been watching my videos for months or even years, but never took for months or even years, but never took for months or even years, but never took the time to click the subscribe button the time to click the subscribe button the time to click the subscribe button below this video. Now, you will see that below this video. Now, you will see that below this video. Now, you will see that we are approaching a very important we are approaching a very important we are approaching a very important milestone on this channel. I never milestone on this channel. I never milestone on this channel. I never thought that I would reach 100,000 thought that I would reach 100,000 thought that I would reach 100,000 subscribers when I started making niche subscribers when I started making niche subscribers when I started making niche DevOps videos. So, we are actually DevOps videos. So, we are actually DevOps videos. So, we are actually getting very close. So, I would like to getting very close. So, I would like to getting very close. So, I would like to ask you, please click the subscribe ask you, please click the subscribe ask you, please click the subscribe button below this video. It is button below this video. It is button below this video. It is completely for free, but it really helps completely for free, but it really helps completely for free, but it really helps my channel with the reach and helping as my channel with the reach and helping as my channel with the reach and helping as many people as possible. If I've taught many people as possible. If I've taught many people as possible. If I've taught you anything over these months or years you anything over these months or years you anything over these months or years that you've been watching my videos, if that you've been watching my videos, if that you've been watching my videos, if I ever gave you some value, please click I ever gave you some value, please click I ever gave you some value, please click the subscribe button. It is completely the subscribe button. It is completely the subscribe button. It is completely for free, but it means so much for a for free, but it means so much for a for free, but it means so much for a small creator like me. Thank you so much small creator like me. Thank you so much small creator like me. Thank you so much for the support and now let's continue. for the support and now let's continue. for the support and now let's continue. So, here's the page of the AWS DevOps So, here's the page of the AWS DevOps So, here's the page of the AWS DevOps agent, but before I dive into this, I agent, but before I dive into this, I agent, but before I dive into this, I want to frame this discussion as want to frame this discussion as want to frame this discussion as follows, which is that AI will never be follows, which is that AI will never be follows, which is that AI will never be worse than it is right now. If there's worse than it is right now. If there's worse than it is right now. If there's something AI cannot do right now, just something AI cannot do right now, just something AI cannot do right now, just give it 12 months and that might change give it 12 months and that might change give it 12 months and that might change a lot. So I don't think that can AI a lot. So I don't think that can AI a lot. So I don't think that can AI replace me today is the right question.

replace me today is the right question. replace me today is the right question. I think the question is really about am I think the question is really about am I think the question is really about am I adapting to this or not? To be fair, I adapting to this or not? To be fair, I adapting to this or not? To be fair, the LLM architecture has its limits. the LLM architecture has its limits. the LLM architecture has its limits. Like you will you see that the Like you will you see that the Like you will you see that the improvements over the past 6 months or improvements over the past 6 months or improvements over the past 6 months or so have been quite marginal relatively. so have been quite marginal relatively. so have been quite marginal relatively. So I think we are reaching the limits of So I think we are reaching the limits of So I think we are reaching the limits of the current architecture, but then it's the current architecture, but then it's the current architecture, but then it's only a matter of time until a new only a matter of time until a new only a matter of time until a new architecture comes out. This is very architecture comes out. This is very architecture comes out. This is very important in any sort of AI discussion important in any sort of AI discussion important in any sort of AI discussion to look at the rate of development. to look at the rate of development. to look at the rate of development. Whatever it cannot do right now, give it Whatever it cannot do right now, give it Whatever it cannot do right now, give it 12 months and probably it will. But that 12 months and probably it will. But that 12 months and probably it will. But that being said, from my observation, the being said, from my observation, the being said, from my observation, the marketing around these AI tools is marketing around these AI tools is marketing around these AI tools is always bigger than reality. And the key always bigger than reality. And the key always bigger than reality. And the key to focus on here is in the verbs. Let me to focus on here is in the verbs. Let me to focus on here is in the verbs. Let me show you what I mean by that. When you show you what I mean by that. When you show you what I mean by that. When you first see this, you will see drive first see this, you will see drive first see this, you will see drive operational excellence with an operational excellence with an operational excellence with an autonomous AI agent that resolves and autonomous AI agent that resolves and autonomous AI agent that resolves and proactively prevents incidents. So the proactively prevents incidents. So the proactively prevents incidents. So the first reaction to this might be, okay, first reaction to this might be, okay, first reaction to this might be, okay, we're done. Like if AI can just solve we're done. Like if AI can just solve we're done. Like if AI can just solve incidents for us, then we're cooked. incidents for us, then we're cooked. incidents for us, then we're cooked. Maybe we need to do some building, but Maybe we need to do some building, but Maybe we need to do some building, but AI handles all of the boring stuff now.

AI handles all of the boring stuff now. AI handles all of the boring stuff now. But this is where you have to start But this is where you have to start But this is where you have to start reading the fine print. It's your always reading the fine print. It's your always reading the fine print. It's your always available operations teammate that available operations teammate that available operations teammate that resolves and proactively prevents resolves and proactively prevents resolves and proactively prevents incidents. It says here that it resolves incidents. It says here that it resolves incidents. It says here that it resolves incidents. But then when you go deeper, incidents. But then when you go deeper, incidents. But then when you go deeper, then we'll see it investigates incidents then we'll see it investigates incidents then we'll see it investigates incidents and identifies operational improvements and identifies operational improvements and identifies operational improvements as an experienced DevOps engineer would, as an experienced DevOps engineer would, as an experienced DevOps engineer would, okay? So that's great. But look closely okay? So that's great. But look closely okay? So that's great. But look closely here what they say. It investigates here what they say. It investigates here what they say. It investigates incidents and identifies operational incidents and identifies operational incidents and identifies operational improvements. Investigates and improvements. Investigates and improvements. Investigates and identifies, but it doesn't solve them at identifies, but it doesn't solve them at identifies, but it doesn't solve them at based on this. What you can do is you based on this. What you can do is you based on this. What you can do is you can ask questions and get instant can ask questions and get instant can ask questions and get instant contextual answers and create custom contextual answers and create custom contextual answers and create custom charts and reports that you can save and charts and reports that you can save and charts and reports that you can save and share with your team. So so far what we share with your team. So so far what we share with your team. So so far what we get is asking questions and creating get is asking questions and creating get is asking questions and creating customer reports that we can share with customer reports that we can share with customer reports that we can share with our team. Let's dive a little bit our team. Let's dive a little bit our team. Let's dive a little bit deeper. Here it says resolve issues deeper. Here it says resolve issues deeper. Here it says resolve issues quickly when they arise. But when we go quickly when they arise. But when we go quickly when they arise. But when we go deeper, what will it say? It's your deeper, what will it say? It's your deeper, what will it say? It's your always on on-call engineer. Cool. It always on on-call engineer. Cool. It always on on-call engineer. Cool. It begins investigating the moment an alert begins investigating the moment an alert begins investigating the moment an alert comes in, which is useful. AWS DevOps comes in, which is useful. AWS DevOps comes in, which is useful. AWS DevOps agent autonomously triages incidents agent autonomously triages incidents agent autonomously triages incidents providing root cause analysis and providing root cause analysis and providing root cause analysis and actions for resolutions. Providing actions for resolutions. Providing actions for resolutions. Providing analysis and actions. Providing actions.

analysis and actions. Providing actions. analysis and actions. Providing actions. Do you see the difference there? It's Do you see the difference there? It's Do you see the difference there? It's not performing actions, it's providing not performing actions, it's providing not performing actions, it's providing actions. It uses its understanding of actions. It uses its understanding of actions. It uses its understanding of your application quickly understand your application quickly understand your application quickly understand dependencies and interactions. It dependencies and interactions. It dependencies and interactions. It streamlines incident response streamlines incident response streamlines incident response automatically routing observations, automatically routing observations, automatically routing observations, finding some mitigation steps through finding some mitigation steps through finding some mitigation steps through your preferred communications channels your preferred communications channels your preferred communications channels such as like ServiceNow and PagerDuty. such as like ServiceNow and PagerDuty. such as like ServiceNow and PagerDuty. So the marketing, the headline is So the marketing, the headline is So the marketing, the headline is resolve issues quickly when they arise, resolve issues quickly when they arise, resolve issues quickly when they arise, but all we get is providing root cause but all we get is providing root cause but all we get is providing root cause analysis and actions for resolutions, analysis and actions for resolutions, analysis and actions for resolutions, which are automatically routed to your which are automatically routed to your which are automatically routed to your preferred communication channels. In preferred communication channels. In preferred communication channels. In other words, it will analyze the other words, it will analyze the other words, it will analyze the situation and send you a Slack message. situation and send you a Slack message. situation and send you a Slack message. That's what we get here. It's not an AI That's what we get here. It's not an AI That's what we get here. It's not an AI agent that is actually solving problems agent that is actually solving problems agent that is actually solving problems and making changes. It takes the and making changes. It takes the and making changes. It takes the information, runs it through a problem information, runs it through a problem information, runs it through a problem and sends it to Slack. That is what you and sends it to Slack. That is what you and sends it to Slack. That is what you get. But let's look deeper. Maybe there get. But let's look deeper. Maybe there get. But let's look deeper. Maybe there is something more hidden in the rest of is something more hidden in the rest of is something more hidden in the rest of the description. Proactively prevent the description. Proactively prevent the description. Proactively prevent future incidents. Well, here's what we future incidents. Well, here's what we future incidents. Well, here's what we get. It provides actionable get. It provides actionable get. It provides actionable recommendations that strengthen these recommendations that strengthen these recommendations that strengthen these four key areas. These are four key areas. These are four key areas. These are recommendations. And this drives recommendations. And this drives recommendations. And this drives continuous improvement without needing continuous improvement without needing continuous improvement without needing to manage a backlog. Wow, that is a to manage a backlog. Wow, that is a to manage a backlog. Wow, that is a massive leap to take here, AWS. You give massive leap to take here, AWS. You give massive leap to take here, AWS. You give us recommendations. That is what they us recommendations. That is what they us recommendations. That is what they give, but you're not proactively give, but you're not proactively give, but you're not proactively preventing future incidents by making preventing future incidents by making preventing future incidents by making changes to my cluster, updating my changes to my cluster, updating my changes to my cluster, updating my application, or maintaining my application, or maintaining my application, or maintaining my infrastructure. It analyzes stuff, runs infrastructure. It analyzes stuff, runs infrastructure. It analyzes stuff, runs it through prompt, and give you a report it through prompt, and give you a report it through prompt, and give you a report or a Slack message. To be fair, it's or a Slack message. To be fair, it's or a Slack message. To be fair, it's useful, but it's not an autonomous agent useful, but it's not an autonomous agent useful, but it's not an autonomous agent that's going to take over our jobs.

that's going to take over our jobs. that's going to take over our jobs. Okay, but the last line then, accelerate Okay, but the last line then, accelerate Okay, but the last line then, accelerate on-demand SRE task handling. Accelerate, on-demand SRE task handling. Accelerate, on-demand SRE task handling. Accelerate, not performing, accelerating. So, what not performing, accelerating. So, what not performing, accelerating. So, what it does, it analyze your application, it does, it analyze your application, it does, it analyze your application, and what you can do is you can query and what you can do is you can query and what you can do is you can query resource health. You can investigate resource health. You can investigate resource health. You can investigate incident patterns. You can track incident patterns. You can track incident patterns. You can track deployments, all in natural language. deployments, all in natural language. deployments, all in natural language. And then, you can create, save, and And then, you can create, save, and And then, you can create, save, and share custom charts and reports that share custom charts and reports that share custom charts and reports that help you track operational metrics and help you track operational metrics and help you track operational metrics and communicate insights with your team. communicate insights with your team. communicate insights with your team. Wow. Well, if that's going to take over Wow. Well, if that's going to take over Wow. Well, if that's going to take over our jobs, then I can sleep pretty good our jobs, then I can sleep pretty good our jobs, then I can sleep pretty good at night. So, the marketing around this at night. So, the marketing around this at night. So, the marketing around this is very clever. The top is drive is very clever. The top is drive is very clever. The top is drive operational excellence with autonomous operational excellence with autonomous operational excellence with autonomous AI agent that resolves and proactively AI agent that resolves and proactively AI agent that resolves and proactively prevents incidents. But, when you read prevents incidents. But, when you read prevents incidents. But, when you read the fine print, it's not resolving, it's the fine print, it's not resolving, it's the fine print, it's not resolving, it's not proactively solving things. It gives not proactively solving things. It gives not proactively solving things. It gives you information. And yes, this is you information. And yes, this is you information. And yes, this is useful, okay? This will impact our jobs useful, okay? This will impact our jobs useful, okay? This will impact our jobs as DevOps engineers. But, this is also as DevOps engineers. But, this is also as DevOps engineers. But, this is also what I saw at KubeCon. I was visiting what I saw at KubeCon. I was visiting what I saw at KubeCon. I was visiting multiple of these stands, and there was multiple of these stands, and there was multiple of these stands, and there was the agentic observability platform, AI the agentic observability platform, AI the agentic observability platform, AI SRE. And when I looked at these demos, SRE. And when I looked at these demos, SRE. And when I looked at these demos, because when I first saw those because when I first saw those because when I first saw those headlines, I was getting a bit worried, headlines, I was getting a bit worried, headlines, I was getting a bit worried, right? I said, "Okay, well, the future right? I said, "Okay, well, the future right? I said, "Okay, well, the future is here. We're cooked." So, I looked at is here. We're cooked." So, I looked at is here. We're cooked." So, I looked at these demos, but what you got was you these demos, but what you got was you these demos, but what you got was you have an application with all your open have an application with all your open have an application with all your open telemetry data, and you could select a telemetry data, and you could select a telemetry data, and you could select a piece of that data, and then the AI piece of that data, and then the AI piece of that data, and then the AI agent, what that meant was you select, agent, what that meant was you select, agent, what that meant was you select, you click push to chat, and then it puts you click push to chat, and then it puts you click push to chat, and then it puts that data into a chat, and then you that data into a chat, and then you that data into a chat, and then you could chat about it with a model. That could chat about it with a model. That could chat about it with a model. That is what they call agentic observability.

is what they call agentic observability. is what they call agentic observability. The marketing around this stuff is very The marketing around this stuff is very The marketing around this stuff is very devious, and it almost fear-mongering, devious, and it almost fear-mongering, devious, and it almost fear-mongering, but it really plays on managers who but it really plays on managers who but it really plays on managers who don't understand the day-to-day work don't understand the day-to-day work don't understand the day-to-day work that we do as DevOps engineers. Like any that we do as DevOps engineers. Like any that we do as DevOps engineers. Like any senior DevOps engineer who has a above senior DevOps engineer who has a above senior DevOps engineer who has a above average understanding of English who average understanding of English who average understanding of English who takes the time to read this will see takes the time to read this will see takes the time to read this will see that this is absolutely trash and almost that this is absolutely trash and almost that this is absolutely trash and almost useless in terms of like actual agentic useless in terms of like actual agentic useless in terms of like actual agentic AI work. This you have to always take AI work. This you have to always take AI work. This you have to always take into account to before you like make a into account to before you like make a into account to before you like make a decision, I'm not going to become a decision, I'm not going to become a decision, I'm not going to become a DevOps engineer because AI can do it. DevOps engineer because AI can do it. DevOps engineer because AI can do it. You really have to read the fine print You really have to read the fine print You really have to read the fine print and understand what these tools are and understand what these tools are and understand what these tools are doing. Okay, taking a step back out of doing. Okay, taking a step back out of doing. Okay, taking a step back out of the emotion because I was ranting a the emotion because I was ranting a the emotion because I was ranting a little bit just there and speaking from little bit just there and speaking from little bit just there and speaking from emotions which is not the most emotions which is not the most emotions which is not the most productive thing to do. I calm down, productive thing to do. I calm down, productive thing to do. I calm down, take a breath, and let's take another take a breath, and let's take another take a breath, and let's take another look at this calmly with a stoic look at this calmly with a stoic look at this calmly with a stoic observing mind and trying to keep observing mind and trying to keep observing mind and trying to keep opinions out of this. So, when I look at opinions out of this. So, when I look at opinions out of this. So, when I look at this from this calmer perspective that I this from this calmer perspective that I this from this calmer perspective that I can put myself into where I do not feel can put myself into where I do not feel can put myself into where I do not feel threatened about my job or anything and threatened about my job or anything and threatened about my job or anything and just thinking about how is this going to just thinking about how is this going to just thinking about how is this going to affect the day-to-day of a DevOps affect the day-to-day of a DevOps affect the day-to-day of a DevOps engineer, you will see that it has a engineer, you will see that it has a engineer, you will see that it has a pretty profound effect. Because right pretty profound effect. Because right pretty profound effect. Because right now, when I am working for a client and now, when I am working for a client and now, when I am working for a client and there's an incident at work, all you get there's an incident at work, all you get there's an incident at work, all you get is there is a problem or the application is there is a problem or the application is there is a problem or the application is down. That is what you get. So, the is down. That is what you get. So, the is down. That is what you get. So, the engineer's task is then to go in and to engineer's task is then to go in and to engineer's task is then to go in and to first take a look at okay, well, the first take a look at okay, well, the first take a look at okay, well, the application is down. Why is it down and application is down. Why is it down and application is down. Why is it down and what went wrong? And that process

what went wrong? And that process what went wrong? And that process already takes manual work. Like when I already takes manual work. Like when I already takes manual work. Like when I was in my first job as a junior DevOps was in my first job as a junior DevOps was in my first job as a junior DevOps engineer, this step literally meant engineer, this step literally meant engineer, this step literally meant SSHing into our servers because we SSHing into our servers because we SSHing into our servers because we didn't have any sort of mature didn't have any sort of mature didn't have any sort of mature observability going on. So, you would observability going on. So, you would observability going on. So, you would literally go into the server and see if literally go into the server and see if literally go into the server and see if the container was still up. Now, things the container was still up. Now, things the container was still up. Now, things have improved since then and I've worked have improved since then and I've worked have improved since then and I've worked in much more mature organizations where in much more mature organizations where in much more mature organizations where you actually have dashboards for this. you actually have dashboards for this. you actually have dashboards for this. So, it becomes easier, but there's So, it becomes easier, but there's So, it becomes easier, but there's always the step of oh, we don't know always the step of oh, we don't know always the step of oh, we don't know what's going on. Let's go and take a what's going on. Let's go and take a what's going on. Let's go and take a look. Now, what these agent tools are look. Now, what these agent tools are look. Now, what these agent tools are going to help with is to already have going to help with is to already have going to help with is to already have done that step for you, okay? So, an done that step for you, okay? So, an done that step for you, okay? So, an alert fires and that will trigger this alert fires and that will trigger this alert fires and that will trigger this AI agent that is going to pull in all of AI agent that is going to pull in all of AI agent that is going to pull in all of the context and depending on where it's the context and depending on where it's the context and depending on where it's running, etc., but it will already be running, etc., but it will already be running, etc., but it will already be able to do this first triaging step for able to do this first triaging step for able to do this first triaging step for you and giving you some recommendations you and giving you some recommendations you and giving you some recommendations of where to look next. Now, this, of where to look next. Now, this, of where to look next. Now, this, honestly, will have a very positive and honestly, will have a very positive and honestly, will have a very positive and big time-saving effect day-to-day big time-saving effect day-to-day big time-saving effect day-to-day reality of a DevOps engineer. It will be reality of a DevOps engineer. It will be reality of a DevOps engineer. It will be super useful because these AI agents, super useful because these AI agents, super useful because these AI agents, they are getting larger and larger they are getting larger and larger they are getting larger and larger context windows and if it can literally context windows and if it can literally context windows and if it can literally see your entire cloud environment and it see your entire cloud environment and it see your entire cloud environment and it can pinpoint might be a networking issue can pinpoint might be a networking issue can pinpoint might be a networking issue or the container is up but it's not or the container is up but it's not or the container is up but it's not connecting to storage, something like connecting to storage, something like connecting to storage, something like that, that will already help a lot and that, that will already help a lot and that, that will already help a lot and you have a place to go. You have a place you have a place to go. You have a place you have a place to go. You have a place to start from. This can literally save to start from. This can literally save to start from. This can literally save two to four hours sometimes because what two to four hours sometimes because what two to four hours sometimes because what can happen as an engineer is that you can happen as an engineer is that you can happen as an engineer is that you start digging and you are looking in the start digging and you are looking in the start digging and you are looking in the complete different direction for where

complete different direction for where complete different direction for where the problem actually is. It has happened the problem actually is. It has happened the problem actually is. It has happened multiple times where I was trying to multiple times where I was trying to multiple times where I was trying to troubleshoot something and I had an troubleshoot something and I had an troubleshoot something and I had an inkling that it was storage-related, but inkling that it was storage-related, but inkling that it was storage-related, but then after half an hour, the 30-minute then after half an hour, the 30-minute then after half an hour, the 30-minute rule, I couldn't figure it out, so I rule, I couldn't figure it out, so I rule, I couldn't figure it out, so I called the help of another engineer who called the help of another engineer who called the help of another engineer who came in with a different perspective and came in with a different perspective and came in with a different perspective and then boom, we found it very quickly. So, then boom, we found it very quickly. So, then boom, we found it very quickly. So, I do think that the agents are going to I do think that the agents are going to I do think that the agents are going to have a very positive impact in this have a very positive impact in this have a very positive impact in this initial phase of the incident. But then initial phase of the incident. But then initial phase of the incident. But then the next step is to actually resolve the next step is to actually resolve the next step is to actually resolve these incidents and this is where DevOps these incidents and this is where DevOps these incidents and this is where DevOps engineers have nothing to fear at this engineers have nothing to fear at this engineers have nothing to fear at this point. Like there's not a serious point. Like there's not a serious point. Like there's not a serious production environment that is going to production environment that is going to production environment that is going to be able to outsource this step to an AI be able to outsource this step to an AI be able to outsource this step to an AI agent. If you have used AI agents to any agent. If you have used AI agents to any agent. If you have used AI agents to any degree, then you will know that the no degree, then you will know that the no degree, then you will know that the no matter how many instructions you give matter how many instructions you give matter how many instructions you give it, it will still sometimes just go it, it will still sometimes just go it, it will still sometimes just go their own way and ignore these kinds of their own way and ignore these kinds of their own way and ignore these kinds of thing. The output of LLMs is thing. The output of LLMs is thing. The output of LLMs is non-deterministic. You can never be sure non-deterministic. You can never be sure non-deterministic. You can never be sure that an output of an LLM is going to be that an output of an LLM is going to be that an output of an LLM is going to be exactly the same from one prompt to the exactly the same from one prompt to the exactly the same from one prompt to the next. And this unpredictability, this next. And this unpredictability, this next. And this unpredictability, this randomness to LLMs because they are randomness to LLMs because they are randomness to LLMs because they are random token generators, means that they random token generators, means that they random token generators, means that they cannot reliably fix these things or cannot reliably fix these things or cannot reliably fix these things or generate output that you can trust on.

generate output that you can trust on. generate output that you can trust on. You will always need a skilled DevOps You will always need a skilled DevOps You will always need a skilled DevOps engineer at the other end to actually engineer at the other end to actually engineer at the other end to actually take this information, process it take this information, process it take this information, process it through its mind and years of through its mind and years of through its mind and years of experience, and then take action experience, and then take action experience, and then take action accordingly. So, a craftsman like me has accordingly. So, a craftsman like me has accordingly. So, a craftsman like me has nothing to fear. A deep craftsman, a nothing to fear. A deep craftsman, a nothing to fear. A deep craftsman, a sovereign craftsman who has honed their sovereign craftsman who has honed their sovereign craftsman who has honed their crafts over 5 plus years have nothing to crafts over 5 plus years have nothing to crafts over 5 plus years have nothing to fear, which is very different, fear, which is very different, fear, which is very different, unfortunately, from software developers. unfortunately, from software developers. unfortunately, from software developers. Like there are different stages. You can Like there are different stages. You can Like there are different stages. You can be a software engineer, but you can also be a software engineer, but you can also be a software engineer, but you can also be a developer whose only skill is to be a developer whose only skill is to be a developer whose only skill is to basically turn a requirement into code. basically turn a requirement into code. basically turn a requirement into code. It's literally just typing code, whereas It's literally just typing code, whereas It's literally just typing code, whereas engineers are sometimes focused on engineers are sometimes focused on engineers are sometimes focused on bigger problems. But the developer, the bigger problems. But the developer, the bigger problems. But the developer, the person who only writes code, the person who only writes code, the person who only writes code, the programmer, those people are very programmer, those people are very programmer, those people are very rapidly being replaced by AI. But I rapidly being replaced by AI. But I rapidly being replaced by AI. But I don't think the same goes for, for don't think the same goes for, for don't think the same goes for, for example, DevOps engineers because we example, DevOps engineers because we example, DevOps engineers because we really need to make these important really need to make these important really need to make these important decisions when we have the root cause decisions when we have the root cause decisions when we have the root cause analysis of the incident. And besides, analysis of the incident. And besides, analysis of the incident. And besides, DevOps engineers have always been about DevOps engineers have always been about DevOps engineers have always been about automating themselves away. At least the automating themselves away. At least the automating themselves away. At least the the serious skilled DevOps engineers the serious skilled DevOps engineers the serious skilled DevOps engineers have always taken this as a goal. I have have always taken this as a goal. I have have always taken this as a goal. I have always taken my goal as a consultant always taken my goal as a consultant always taken my goal as a consultant doing Kubernetes consulting for large doing Kubernetes consulting for large doing Kubernetes consulting for large companies. I've always seen my role as companies. I've always seen my role as companies. I've always seen my role as making myself obsolete as quickly as making myself obsolete as quickly as making myself obsolete as quickly as possible by building automated systems possible by building automated systems possible by building automated systems that people can trust on. This is why we that people can trust on. This is why we that people can trust on. This is why we use Kubernetes with reconciliation use Kubernetes with reconciliation use Kubernetes with reconciliation loops, GitOps, automated testing and

loops, GitOps, automated testing and loops, GitOps, automated testing and deployment pipelines. Every single layer deployment pipelines. Every single layer deployment pipelines. Every single layer of the DevOps engineer's profession is of the DevOps engineer's profession is of the DevOps engineer's profession is about taking manual tasks and automating about taking manual tasks and automating about taking manual tasks and automating them. And AI is just going to be the them. And AI is just going to be the them. And AI is just going to be the next layer of this. It's going to be the next layer of this. It's going to be the next layer of this. It's going to be the next iterations of this. So, a serious next iterations of this. So, a serious next iterations of this. So, a serious craftsman is just going to take this and craftsman is just going to take this and craftsman is just going to take this and roll with it and adapt and get even more roll with it and adapt and get even more roll with it and adapt and get even more work as a result. And I'm going to work as a result. And I'm going to work as a result. And I'm going to explain why in a second. If you are a explain why in a second. If you are a explain why in a second. If you are a craftsman, and if you care about your craftsman, and if you care about your craftsman, and if you care about your work, and if you have a deep level of work, and if you have a deep level of work, and if you have a deep level of skill and knowledge, there is truly skill and knowledge, there is truly skill and knowledge, there is truly nothing to fear here. The person who is nothing to fear here. The person who is nothing to fear here. The person who is at risk because of these AI tools are at risk because of these AI tools are at risk because of these AI tools are DevOps engineers who are clicking around DevOps engineers who are clicking around DevOps engineers who are clicking around in the AWS portal, who when they get an in the AWS portal, who when they get an in the AWS portal, who when they get an incident, they all they do is they copy incident, they all they do is they copy incident, they all they do is they copy it, put it into Google, and hope there it, put it into Google, and hope there it, put it into Google, and hope there is a Stack Overflow post about it, or is a Stack Overflow post about it, or is a Stack Overflow post about it, or ask ChatGPT without actually ask ChatGPT without actually ask ChatGPT without actually understanding what they're doing. Yeah, understanding what they're doing. Yeah, understanding what they're doing. Yeah, these guys are probably at risk by these these guys are probably at risk by these these guys are probably at risk by these AI tools. But the person who actually AI tools. But the person who actually AI tools. But the person who actually understands what they're doing with a understands what they're doing with a understands what they're doing with a deep level of skill, who would approach deep level of skill, who would approach deep level of skill, who would approach his craft like a craftsman, they have his craft like a craftsman, they have his craft like a craftsman, they have nothing to fear. So, if you're watching nothing to fear. So, if you're watching nothing to fear. So, if you're watching this and you want to become a DevOps this and you want to become a DevOps this and you want to become a DevOps craftsman like me, then check out the craftsman like me, then check out the craftsman like me, then check out the free DevOps career blueprint, which is free DevOps career blueprint, which is free DevOps career blueprint, which is in the description below. It is in the description below. It is in the description below. It is completely for free. It contains over 20 completely for free. It contains over 20 completely for free. It contains over 20 hours of free courses that I'm giving to hours of free courses that I'm giving to hours of free courses that I'm giving to you for free. It gives my entire you for free. It gives my entire you for free. It gives my entire approach to becoming a DevOps engineer approach to becoming a DevOps engineer approach to becoming a DevOps engineer at the level of a craftsman who will at the level of a craftsman who will at the level of a craftsman who will have nothing to fear from these AI have nothing to fear from these AI have nothing to fear from these AI tools. It's completely for free. Check tools. It's completely for free. Check tools. It's completely for free. Check it out. The link is in the description, it out. The link is in the description, it out. The link is in the description, and let me know what you think. So, is and let me know what you think. So, is and let me know what you think. So, is this going to lead to fewer DevOps jobs?

this going to lead to fewer DevOps jobs? this going to lead to fewer DevOps jobs? And the answer is yes and no. Yes, it's And the answer is yes and no. Yes, it's And the answer is yes and no. Yes, it's going to lead to fewer DevOps jobs going to lead to fewer DevOps jobs going to lead to fewer DevOps jobs because if you have a large enterprise because if you have a large enterprise because if you have a large enterprise organization with a team of 10 plus organization with a team of 10 plus organization with a team of 10 plus DevOps engineers, I think tools like DevOps engineers, I think tools like DevOps engineers, I think tools like this are going to maybe cost a two jobs this are going to maybe cost a two jobs this are going to maybe cost a two jobs in that team. But this is not in that team. But this is not in that team. But this is not necessarily because of the effectiveness necessarily because of the effectiveness necessarily because of the effectiveness of this tool. The thing is that in large of this tool. The thing is that in large of this tool. The thing is that in large environments like this with 10 plus environments like this with 10 plus environments like this with 10 plus engineers in a team, there's always a engineers in a team, there's always a engineers in a team, there's always a couple of guys who actually only work couple of guys who actually only work couple of guys who actually only work for one or two hours a day. They are for one or two hours a day. They are for one or two hours a day. They are just flying under the radar. They are just flying under the radar. They are just flying under the radar. They are doing the bare minimum not to get fired doing the bare minimum not to get fired doing the bare minimum not to get fired and they collect a six-figure salary for and they collect a six-figure salary for and they collect a six-figure salary for doing that. Every big environment has a doing that. Every big environment has a doing that. Every big environment has a couple of guys who are doing this. With couple of guys who are doing this. With couple of guys who are doing this. With the advance of these tools, then there the advance of these tools, then there the advance of these tools, then there is not the incident response facade to is not the incident response facade to is not the incident response facade to hide behind. You cannot say, "Oh, yeah, hide behind. You cannot say, "Oh, yeah, hide behind. You cannot say, "Oh, yeah, I was working on this incident all day I was working on this incident all day I was working on this incident all day yesterday." No, because now you can see, yesterday." No, because now you can see, yesterday." No, because now you can see, well, the AI told you what the problem well, the AI told you what the problem well, the AI told you what the problem was. Why did it take you so long to fix was. Why did it take you so long to fix was. Why did it take you so long to fix it? So, what this is going to lead to, I it? So, what this is going to lead to, I it? So, what this is going to lead to, I think, is what I call trimming the fat.

think, is what I call trimming the fat. think, is what I call trimming the fat. So, it's going to be trimming away these So, it's going to be trimming away these So, it's going to be trimming away these kinds of guys who are doing the bare kinds of guys who are doing the bare kinds of guys who are doing the bare minimum and it's going to make DevOps minimum and it's going to make DevOps minimum and it's going to make DevOps engineers a lot more efficient in terms engineers a lot more efficient in terms engineers a lot more efficient in terms of incident response. It's going to lead of incident response. It's going to lead of incident response. It's going to lead a to a reduction of work hours, but a to a reduction of work hours, but a to a reduction of work hours, but historically, from my experience, any historically, from my experience, any historically, from my experience, any team that I have always worked in is team that I have always worked in is team that I have always worked in is always overloaded. So, it's not always overloaded. So, it's not always overloaded. So, it's not necessarily that there is less work to necessarily that there is less work to necessarily that there is less work to do. It's rather that, "Oh, now we have do. It's rather that, "Oh, now we have do. It's rather that, "Oh, now we have time to do all of this accumulated tech time to do all of this accumulated tech time to do all of this accumulated tech debt that we can resolve or now we can debt that we can resolve or now we can debt that we can resolve or now we can finally start upgrading this old finally start upgrading this old finally start upgrading this old infrastructure that we have been infrastructure that we have been infrastructure that we have been postponing all this time." Yes, it's postponing all this time." Yes, it's postponing all this time." Yes, it's going to lead to some trimming of the going to lead to some trimming of the going to lead to some trimming of the fat. It's going to save some hours for fat. It's going to save some hours for fat. It's going to save some hours for DevOps engineers at this stage, but I DevOps engineers at this stage, but I DevOps engineers at this stage, but I don't think this is going to lead to don't think this is going to lead to don't think this is going to lead to mass layoffs by or DevOps engineers just mass layoffs by or DevOps engineers just mass layoffs by or DevOps engineers just yet. Now, the second part to this yet. Now, the second part to this yet. Now, the second part to this question, are we going to lose our jobs question, are we going to lose our jobs question, are we going to lose our jobs as DevOps engineers, is the following. as DevOps engineers, is the following. as DevOps engineers, is the following. You see, if you are a modern DevOps You see, if you are a modern DevOps You see, if you are a modern DevOps engineer, you have a skill that is engineer, you have a skill that is engineer, you have a skill that is actually responsible for running these actually responsible for running these actually responsible for running these AI tools and AI agent, namely, AI tools and AI agent, namely, AI tools and AI agent, namely, Kubernetes. Think about it. ChatGPT runs Kubernetes. Think about it. ChatGPT runs Kubernetes. Think about it. ChatGPT runs on Kubernetes. Anthropic Claude runs on on Kubernetes. Anthropic Claude runs on on Kubernetes. Anthropic Claude runs on Kubernetes. Google Gemini, XAI, Grok, Kubernetes. Google Gemini, XAI, Grok, Kubernetes. Google Gemini, XAI, Grok, all runs on Kubernetes. This is public all runs on Kubernetes. This is public all runs on Kubernetes. This is public information that you can look up right information that you can look up right information that you can look up right now. These AI models are being trained, now. These AI models are being trained, now. These AI models are being trained, built, and run on Kubernetes. So, if all built, and run on Kubernetes. So, if all built, and run on Kubernetes. So, if all of these AI tools are growing and all of of these AI tools are growing and all of of these AI tools are growing and all of these use cases that are going to be these use cases that are going to be these use cases that are going to be replacing knowledge workers, frankly, in replacing knowledge workers, frankly, in replacing knowledge workers, frankly, in the next decade or so, we control the the next decade or so, we control the the next decade or so, we control the infrastructure that all of this is

infrastructure that all of this is infrastructure that all of this is running on. So, if AI grows, our jobs running on. So, if AI grows, our jobs running on. So, if AI grows, our jobs also will grow. And this is not my also will grow. And this is not my also will grow. And this is not my opinion, the CNCF just released a state opinion, the CNCF just released a state opinion, the CNCF just released a state of the cloud native development report of the cloud native development report of the cloud native development report where they say that 66% of organizations where they say that 66% of organizations where they say that 66% of organizations run generative AI workloads on run generative AI workloads on run generative AI workloads on Kubernetes. And quoting the report Kubernetes. And quoting the report Kubernetes. And quoting the report directly, Kubernetes is the de facto directly, Kubernetes is the de facto directly, Kubernetes is the de facto operating system for AI. And if you want operating system for AI. And if you want operating system for AI. And if you want to have the deep dive of that report, to have the deep dive of that report, to have the deep dive of that report, you can check out the video over here. you can check out the video over here. you can check out the video over here. So, McKinsey recently made an estimate So, McKinsey recently made an estimate So, McKinsey recently made an estimate that over $7 trillion that over $7 trillion that over $7 trillion are going to be invested in AI data are going to be invested in AI data are going to be invested in AI data centers by 2030. And all of these AI centers by 2030. And all of these AI centers by 2030. And all of these AI data data centers are going to be data data centers are going to be data data centers are going to be running Kubernetes clusters, and they're running Kubernetes clusters, and they're running Kubernetes clusters, and they're going to be needing people to operate going to be needing people to operate going to be needing people to operate those Kubernetes clusters. So, this is those Kubernetes clusters. So, this is those Kubernetes clusters. So, this is why I think Kubernetes engineers why I think Kubernetes engineers why I think Kubernetes engineers specifically are pretty much safe from specifically are pretty much safe from specifically are pretty much safe from AI replacement at this stage. Because, AI replacement at this stage. Because, AI replacement at this stage. Because, yes, maybe these annoying boring tasks yes, maybe these annoying boring tasks yes, maybe these annoying boring tasks like initial incident response are going like initial incident response are going like initial incident response are going to be reduced, but that means that to be reduced, but that means that to be reduced, but that means that there's even more time for us to start there's even more time for us to start there's even more time for us to start building these AI infrastructure layers, building these AI infrastructure layers, building these AI infrastructure layers, for example, right? So, if you stick for example, right? So, if you stick for example, right? So, if you stick with the times and if you start with the times and if you start with the times and if you start mastering these tools and adopting them, mastering these tools and adopting them, mastering these tools and adopting them, then I think AI is only going to then I think AI is only going to then I think AI is only going to accelerate our careers as DevOps accelerate our careers as DevOps accelerate our careers as DevOps engineers. And I think there's just engineers. And I think there's just engineers. And I think there's just going to be more jobs for us in the going to be more jobs for us in the going to be more jobs for us in the future and time to come. And especially future and time to come. And especially future and time to come. And especially six-figure consulting opportunities. I'm six-figure consulting opportunities. I'm six-figure consulting opportunities. I'm like super happy with what's happening like super happy with what's happening like super happy with what's happening here because I can only see here because I can only see here because I can only see opportunities in the future for someone opportunities in the future for someone opportunities in the future for someone who has a skill set like mine. So, the who has a skill set like mine. So, the who has a skill set like mine. So, the question isn't really about is AI going

question isn't really about is AI going question isn't really about is AI going to replace me, the question is am I to replace me, the question is am I to replace me, the question is am I learning the tools and infrastructure learning the tools and infrastructure learning the tools and infrastructure that this AI is running on. So, if the that this AI is running on. So, if the that this AI is running on. So, if the answer to that is yes, then you have answer to that is yes, then you have answer to that is yes, then you have nothing to fear. So, the DevOps career nothing to fear. So, the DevOps career nothing to fear. So, the DevOps career blueprint that you can get completely blueprint that you can get completely blueprint that you can get completely for free focuses on this. It will teach for free focuses on this. It will teach for free focuses on this. It will teach you all of the skills you need to you all of the skills you need to you all of the skills you need to prevent yourself from being replaced prevent yourself from being replaced prevent yourself from being replaced this way by focusing on the right tools. this way by focusing on the right tools. this way by focusing on the right tools. The link is in the description below. The link is in the description below. The link is in the description below. It's completely for free, so check it It's completely for free, so check it It's completely for free, so check it out. So, yes, the AWS DevOps agent, it out. So, yes, the AWS DevOps agent, it out. So, yes, the AWS DevOps agent, it looks like a useful tool, but ignore all looks like a useful tool, but ignore all looks like a useful tool, but ignore all of the videos that are going to come of the videos that are going to come of the videos that are going to come where they are going to say that DevOps where they are going to say that DevOps where they are going to say that DevOps is over. If you truly have industry is over. If you truly have industry is over. If you truly have industry knowledge, you know what's going on. We knowledge, you know what's going on. We knowledge, you know what's going on. We control the infrastructure that these control the infrastructure that these control the infrastructure that these tools are running on, so we have nothing tools are running on, so we have nothing tools are running on, so we have nothing to fear. Learn the infrastructure to fear. Learn the infrastructure to fear. Learn the infrastructure deeply, operate the clusters, and you deeply, operate the clusters, and you deeply, operate the clusters, and you will go far. Let me know in the comments will go far. Let me know in the comments will go far. Let me know in the comments if you agree. Thank you so much for if you agree. Thank you so much for if you agree. Thank you so much for watching, and I'll see you in the next watching, and I'll see you in the next watching, and I'll see you in the next video.