How AI is Changing DevOps Careers | What You Need to Know

0:00Even if you paid the most money, the hundreds of dollars for the high-end

0:04Claude models, high-end GPT models, you are at best gonna get about 70%

0:10accuracy for a software developer.

0:1270% means it won't work, it's not gonna function at 70%.

0:15Correct.

0:16The app will not function.

0:17It was a weird disconnect between what I was hearing on the internet and what

0:20even some AI conference talks were talking about, and then what everyone

0:25I knew and everyone I would talk to.

0:27The AI can't be fired.

0:29The company isn't going to fire anthropic because it deployed a Kubernetes

0:33cluster with the API for Kubernetes, accidentally public on the internet,

0:37the person who committed the AI code is gonna get fired, so there's still

0:41gotta be a human that has to go.

0:43I'm willing to take this risk.

0:45I'm willing to push that deploy button.

0:50I spent two hours talking with Brett Fisher about AI and DevOps, and what he

0:56told me completely changed how I think about the combination of these two.

1:01Here's the most common question everyone wants answered.

1:04Is AI going to automate DevOps work and make engineers obsolete, or are

1:10we still many years away from that?

1:13Will AI do our work so well that we will need.

1:15Fewer engineers should we really believe when people say engineers who know how

1:20to use AI are becoming more valuable and not obsolete, should we believe the hype?

1:25Or is there still massive gap between what AI companies promise and what's

1:30actually happening in production?

1:32And that's exactly what I discussed with Brett to address these questions with

1:37actual data and real world experience.

1:40Not just personal opinions and theory.

1:42Brett has been teaching Docker and DevOps for over a decade, but

1:47he has been diving deep into the AI tools, concepts, and trends.

1:52He also spent two years walking around conferences asking engineers about

1:57their real practical experiences with different AI tools, and all this showed

2:04him the truth about AI adoption in it.

2:08And DevOps specifically.

2:09He even launched a podcast specifically about AI for DevOps.

2:14So he has some real, tangible advice and insights on this topic, which you

2:20will get fully in this conversation.

2:23So let's dive in.

2:29The first thing I wanted to understand from Brett was simple.

2:33Can AI actually do DevOps work?

2:36Everyone is talking about AI automating infrastructure, AI writing, Terraform

2:40code, ai, managing Kubernetes clusters.

2:43So I asked him directly, is this real or is this just a hype?

2:49I saw that you are.

2:51Creating a lot of videos about AI in DevOps or agentic DevOps.

2:55So for someone who has been doing DevOps and who has been doing a little

3:00bit of like software development or IT infrastructure, who hasn't

3:04touched the AI topic yet, how would you describe that ecosystem?

3:08Like how messy it is, how chaotic it is, how difficult it is to choose the right

3:13tooling, or what do you focus on instead?

3:15Like tools or the use cases?

3:17How would you describe that?

3:18Only eight months ago, right where we, I wasn't even aware of the

3:22agentic idea of how we were going to really automate a LLM in a loop.

3:29Essentially, that's what most of these agents are, is just fancy logic in a loop.

3:32That was very much a surprise to me.

3:34I mean, I was like everyone else.

3:36I was using chat, GPTI was using cursor and copilot and trying all

3:40the different tools, trying to figure out what was useful for

3:43me, but also trying to casually.

3:47On my spare time, maybe spend a little time figuring out

3:49what other people are doing.

3:50But in 2024, even very early 2025, no one in DevOps was really talking

3:55about using AI tools to do their job.

3:58And other than maybe writing YAML with an AI that you would just, you know,

4:03one shot it, that's where you just ask it a question and you take the

4:05response and there's no agent involved.

4:07You mentioned that you dug deeper, you know, you kind of had this realization,

4:11okay, there must be something there.

4:13There's.

4:13Scold.

4:14From the information that you got, I'm pretty sure that you started learning and

4:18researching with certain expectations.

4:20Right?

4:20So you had some kind of assumption probably in your head, like you didn't

4:23know what you were gonna discover.

4:25What was something that you discovered during this search that you didn't

4:28expect in terms of like, I didn't expect it to work this well.

4:32Like to be that mature and also the opposite.

4:35When you dug deeper, like, I didn't expect it to be this bad actually,

4:40you know, compared to the hype.

4:41Like what were some of the details or like aha moments that you had during

4:45that three month of intense research?

4:48Yeah, great question.

4:49Most of it is not good news, but that like maybe a, a surprise on

4:53the good side was when you can get one of the high-end frontier models.

4:58In an agent loop and it actually spits out something that you know how to read.

5:03Like the, the key is that right now AI is really aren't gonna

5:06do anything of your job for you.

5:07You, you've gotta babysit 'em.

5:08But when it gets it right and you get, you know, whether it's a, a couple

5:11hundred lines of YAML or, uh, a whole new.

5:14Class in a Java file or a JavaScript file, you get a whole TypeScript perfect

5:18layout that does feel like magic, right?

5:21When you just saved yourself 20, 30 minutes of research.

5:24And I've been making a lot of CLI tools lately because this stuff is so easy.

5:27So tools to automate my work, tools to help give DevOps examples of tools, and I

5:31like Golan as my that's go is my preferred language for, for DevOps tooling.

5:36When it gets it right, which is rare and it writes the way I would and

5:40stuff like that, it feels so exciting and it feels like I'm a kid again.

5:44Discovering code, discovering how to automate things and play with things,

5:49that just doesn't happen often enough yet.

5:51And even the best models that we have, the absolute best models that the.

5:55Top of the rankings are only going to fix about two thirds, maybe, if you're lucky.

6:00Three fourths of GitHub issues.

6:02We have this thing called Suite Bench now, and all the models are listed

6:06there and they're constantly competing with each other for the top rankings.

6:09But even if you paid.

6:11The most money, the hundreds of dollars for the high-end clawed

6:14models or the high-end GPT models, you are at best gonna get about

6:1970% accuracy, which for a software developer, 70% means it won't work.

6:23It's not gonna function at 70%.

6:25Correct.

6:25The app will not function eventually.

6:26There'll be so many bugs.

6:27Right.

6:28Let me explain what Sweep Bench is and why this.

6:3170% number matters so much.

6:35Sweep Bench is a benchmark that tests how well AI models can solve real GitHub

6:43issues from actual open source projects.

6:46So these are not some theoretical problems.

6:49These are actual bugs and features that real developers submitted.

6:55In those projects and the best models like Claude Sonnet, G, PT four, the

7:01most expensive, most advanced AI available out there, they solve about

7:0870% of these issues correctly now.

7:1170% sounds pretty good.

7:14Right?

7:14But here's the problem.

7:15In DevOps specifically, partial correctness does not work because if your

7:22Terraform configuration is 70% correct, your infrastructure will not deploy.

7:28If your Kubernetes YAML is 70% correct, your pods will not start if your

7:33CSD pipeline is just 70% correct.

7:36It will fail, and you will need to troubleshoot and identify what the

7:39issues are, what things are breaking.

7:41So you will need to understand how the pipeline, the manifest file,

7:46and the Terraform code work because you'll need to fix those issues.

7:50And this is fundamentally different from other tasks where AI works well.

7:56Because if AI writes a blog post, that's 70% correct, you

8:02can edit the other 30%, right?

8:05But infrastructure code does not work that way.

8:07It's binary.

8:08Either it works or it doesn't.

8:10And this is why you can't just prompt an AI and walk away.

8:14You need to understand what it's building.

8:17You need to verify the output, and you need the knowledge to spot the 30%.

8:23That's wrong.

8:24I think there is a moment with every developer where they're picking up

8:27steam and then they realize that we're still many years, I think away from

8:32these models being so trustworthy that we can just pass on our requests

8:37and not have to look at the answers.

8:39So the idea of like an AI writing, the PR, and then an AI reviewing

8:43the PR and approving the pr.

8:45Whether it's for software code or develop or DevOps, yaml, I think we're still

8:49many years away from that for sort of fully ai, which is why I'm not worried

8:53about jobs and and, and stuff like that.

8:55I think most of that stuff is really hyped up and don't believe anything in ai.

9:00If they work for an AI company, you just can't trust 'em because they're living.

9:04They're not living in reality with the rest of us.

9:06They're living like 10 years in the future, or 15 years in the future.

9:09And

9:10so I have, uh, three follow up questions on those.

9:12So first of all, because there is a difference between tasks

9:16that software engineers are doing or software developers mm-hmm.

9:18And the tasks that DevOps engineers are doing there, you

9:21know, writing code is pretty much different from, you know, your.

9:26Creating and designing an architecture, and then you are kind of putting those

9:30Lego boxes and, and pieces together to build this entire thing, right?

9:34So there is less writing code, but more like mixing and matching,

9:37matching different tools.

9:38So how would you compare?

9:40So is.

9:41AI usable on top of DevOps because one thing that the immediately

9:47stood out for me was that DevOps was always about automation, right?

9:51Right.

9:51So we had lots of bottlenecks.

9:52We had human in the loop, you know, things that slowed down the process.

9:55Mostly it was manual work and DevOps was like, let's automate and, and

9:59remove those bottlenecks and humans.

10:01Where it's not necessary.

10:02And AI conceptually, like agentic AI conceptually does the same, which

10:06is let's automate things which are tedious, let's make it more intelligent.

10:10So it's like intelligent automation versus like DevOps automation.

10:14So is there any value in edit using AI on top of DevOps, or would you say

10:20it's kind of not really like valuable as much as for software development?

10:25Like what would you say?

10:26I think both are very valuable.

10:28Whether if you're gonna compare like software development versus DevOps

10:31engineering, I would say that they're very much, if we're going with just general

10:35accuracy of 70%, if that's a ballpark, I think they're both very much usable.

10:39I think the difference is, is that DevOps, it's not as intuitive because

10:43we don't necessarily have a tool sitting in front of us that's gonna explain

10:47exactly where and how we we're supposed to automate with AI today, right?

10:51Like all of the.

10:52All the IDs are all focused on code generation.

10:55They happen to generate YAML just as well, or ML or whatever

10:59the Js ON, or whatever you need.

11:00They do that just as well as they do the rest of software engineering.

11:03But we don't have, I wouldn't say we have a lot of, especially open source,

11:08we have very little DevOps tooling and open source that lets us just go.

11:12So I think the real challenge for any DevOps person today who's

11:15trying to adopt AI is where is it working and where is it not?

11:18And where are the patterns that I can just, you know, where are the recipes?

11:21I need recipes.

11:22You know, I need cookbooks just to be able to automate something in GitHub.

11:26Or can I use a W S'S MCP server to spin up Kubernetes clusters

11:32with very little effort?

11:34The answer is yes, absolutely.

11:36The real trick is how do we narrow that scope down per task to give

11:39that AI a very, you know, guardrails.

11:42To keep it safe, to make sure that humans are reviewing it.

11:45I feel like we can bring in AI to DevOps today as long as it's only

11:48doing one little part of the job.

11:50And then we slowly, like in everything in DevOps, we slowly expand.

11:54Our new automation.

11:55I mean, this is no different than if you want to implement a

11:57new CI platform for your team.

11:59Maybe you wanna move from Jenkins to GitHub.

12:01You're not gonna do that all in a day, right?

12:03That's gonna be too crazy risky.

12:05This is really important.

12:07Software developers have tools like Cursor, GitHub, co-pilot, visual

12:11Studio code extensions, all designed specifically for writing code with ai.

12:16You open your IDE and AI is right there helping you.

12:20But for DevOps engineers.

12:21The tooling landscape is much more scattered.

12:25You're not just writing code as a DevOps engineer.

12:27You are configuring infrastructure across multiple cloud providers.

12:31You are setting up CICD pipelines maybe in different platforms,

12:35managing COR clusters, writing terraform configurations, setting

12:39up monitoring and alerting in the cluster or on a cloud platform,

12:43managing secrets and security policies.

12:45And each of these lives in a different tool or different

12:48platform, different interface.

12:50There is no single AI assistant that understands your entire DevOps workflow.

12:55So the challenge is not can AI help with DevOps?

12:59The challenge is, where do I even start?

13:02And Brett's answer is clear.

13:04Start small and specific.

13:07Pick one task.

13:08Automate that with ai.

13:10Make sure it works and then expand.

13:13Do not try to automate your entire infrastructure with AI on day one

13:18because that's how you break production.

13:21So instead just pick something narrow and more specific and give

13:26your AI clear guardrails to protect from messing up anything important.

13:33And at the beginning, keep humans in the review loop.

13:37And once you have that fully working, then expand slowly.

13:41And this is exactly how we adopted every other DevOps tool.

13:46When Kubernetes came out, you didn't immediately migrate all your

13:50applications to Kubernetes, right?

13:51You started with one service or one application.

13:54Maybe you did a test run, you made sure it worked and then migrated more.

13:59AI is the same.

14:00It's a new tool in your toolkit, so you introduce it and you

14:05migrate your stuff to it.

14:06Incrementally,

14:14so we know the limitation.

14:15But what are the practical workflows where AI is useful right now?

14:22Not in the future, not someday, but today.

14:25And Brett's answer was surprisingly specific and it starts with CI ICD.

14:30So if you're experimenting with that, absolutely.

14:32I think that CI is a wonderful place to play right now because it's not yet

14:37building production deployments for you, which also is an area that you can

14:40experiment with an AWS engineer, Raj.

14:43Who showed us how to use the MCP servers at AWS to build Kubernetes cluster.

14:47And it is possible, it is totally doable.

14:49It built cloud formation, it built Terraform for us can use CDK.

14:54You can use any of these ways to program the infrastructure and try it out,

14:57but you're wanting, you're what you're gonna realize through your process.

15:01I think if we're gonna imagine anyone out here that's gonna play

15:04with this stuff after listening to us, that you're gonna realize that.

15:07This is all going to, the more you try to automate it with ai, the more

15:11it's gonna fo focus on you having to document, document, document and

15:15give that documentation to a bigger, increasing context to your ai.

15:19So it ends up having all the documentation, all your standards.

15:23What are your requirements?

15:24There are specific IAM groups that has to use, you're gonna have to

15:27give it all that crazy detail.

15:29Now let me explain why CICD is the perfect starting point for AI in DevOps.

15:36And what MCP actually means first.

15:39C, C, D.

15:40Why is this the best place to start?

15:42One.

15:43CICD pipelines are relatively isolated.

15:46If your AI generates a bad pipeline, it fails to build, but it doesn't

15:51take down the production environment.

15:54So the risk radius is actually small.

15:57Two pipeline code is repetitive.

16:00You are doing similar tasks across different projects.

16:03You run tests, build container images.

16:06Push them to the registry, and then you deploy that newly built

16:10image to deployment environment.

16:12And AI is good at repetitive patterns.

16:16Three, you can iterate quickly.

16:19If the AI generated pipeline does not work, you fix it and you try

16:23it again, no production impact.

16:25And four, the feedback loop is fast.

16:29You commit the pipeline, it runs.

16:31You see if it works or not, and within minutes, you basically

16:34know if the AI got it right.

16:36Now, MCP, which stands for Model context Protocol, and this is

16:41becoming really important in the AI landscape, in very simple words.

16:45MCP is basically a way to give AI access to your tools and systems through APIs.

16:53So in the AWS example that Brett mentioned, the AI can

16:57actually talk to AWS APIs.

17:00It can create infrastructure, it can build current clusters, it can write terraform

17:04or cloud formation code and so on.

17:07But here is the critical insight.

17:09The more you automate with ai, the more documentation you need.

17:14Why?

17:15Because AI needs context.

17:18It needs to know your naming conventions, which IAM roles to use your security

17:24requirements and guidelines, which regions you are deploying to your

17:29tagging standards, or how you structure your infrastructure code.

17:33And all of that context needs to be documented and fed to the AI because

17:39otherwise it'll make assumptions and those assumptions will most probably be wrong

17:45and.

17:46That's gonna actually improve your project.

17:48There's a conversation we're all having right now that AI might

17:51actually save the testing slash qa slash documentation people because

17:57they are experts in how all that stuff needs to happen and AI needs that.

18:01Needs that level of context in order to help us create reliable

18:05infrastructure, time and time and time again, and reliable automation.

18:09I think this is actually gonna be helpful for us.

18:11I really don't think it's gonna replace a lot of us.

18:13It's just gonna make us go bigger.

18:15The story.

18:17I always tell is because I'm a gray beard and I've been around for over 30 years

18:21in tech, that I have seen this wave.

18:23This one's happening faster than every wave I've seen, but in the

18:26nineties we went to from mainframe to pc, and the early two thousands we

18:30went from hardware to virtualization.

18:32Five years later, we went from on-prem to cloud.

18:35Then we went five years later from cloud to containers, and now we're just

18:40automating all of that and in each level.

18:43Every single time we went through those phases.

18:45A CIS admin, we didn't really always have the DevOps term, but a CIS admin or

18:49someone who's ops, someone who's cares about infrastructure and deployments

18:53and automation and management.

18:54So in all those generations of us moving from one evolution of

18:59infrastructure, or whether it was a PC or servers or whatever, each time.

19:03The CIS admin grew in nineties.

19:05I could only really manage 10 servers.

19:07I don't even think I had 10.

19:09I had to babysit them.

19:10Everything was manual.

19:11There was no scripting or automation.

19:13Open source wasn't really popular yet in terms of typical enterprise

19:16infrastructure and as it open source grew as the tooling grew over those decades.

19:21And I have a chart that shows like once it's admin to 10 servers.

19:24Then with virtualization, it was one CIS admin with a hundred servers.

19:27Then with the cloud it was one CIS admin with a thousand servers.

19:30Then with containers it, it sort of went beyond servers and you thought

19:33about workloads and it went to one CIS admin or DevOps person could do 10,000

19:37containers, and our tooling got better, our automation got better, and each time

19:42it was all about a single individual managing a bigger and bigger fleet.

19:46So I don't look at that as DevOps lost jobs or in ops, lost jobs.

19:51We just were able to manage more and it turns out.

19:53So far, at least in my whole career, that companies and

19:57organizations, they will take as much infrastructure as you can give them.

20:00They will take as much automation.

20:02They always wanna run more stuff.

20:04I've never seen a team have an nq.

20:06Their ticket queue be zero forever.

20:09Everyone's got broke, broken stuff.

20:11Everyone needs more, more things.

20:12They wanna launch more apps, more copies of the app to make it stable.

20:15More for capacity, more disaster recovery locations where we can spin it up on

20:20the fly in case this place goes down.

20:22Like that's been.

20:23Expanding and expanding.

20:24Expanding, and I don't think AI is going to cause that to stop.

20:26I think AI is just gonna cause this all to manage more.

20:29Now, a quick note here.

20:31While everyone's talking about ai, just go and check any DevOps job postings.

20:37Right now they're asking for Kubernetes, Terraform, Docker, CICD.

20:43You will not find must know how to Useche G. PT on DevOps, job descriptions

20:49or cloud engineer job descriptions.

20:51I actually analyzed over a hundred.

20:54DevOps job posts from dozens of different countries to see what companies are

21:00actually hiring for, what skills they're looking for, regardless of

21:04the role, whether it's cloud engineer, platform engineer, DevOps engineer,

21:08whatever related to DevOps technologies.

21:11And I'll link that video here if you want to see the breakdown.

21:14And Brett confirmed this as well in the interview, you still need DevOps skills.

21:19They are the foundational knowledge that AI does not work without.

21:24But as we all know, DevOps ecosystem is pretty overwhelming.

21:28I think it's extremely important.

21:30To have a clear guide for how to navigate this ecosystem, and that's why we created

21:37DevOps Starter Kit that shows you exactly which skills to learn in which sequence,

21:45based on what actually gets people hired.

21:48So instead of getting caught up in this AI hype, focus on skills that are

21:54demanded right now on job market the most.

21:56So be sure to grab it below.

21:58Link is gonna be in the description because you need to learn DevOps first

22:03and you can add AI on top of it once you have properly mastered DevOps.

22:09Now this historical perspective is super important.

22:13You can literally take notes on this.

22:15So one engineer managing 10 physical servers became a hundred virtual servers,

22:23then thousand virtual servers, then 10,000 containers Each automation technology did

22:29not reduce the number of engineers needed.

22:32It actually increased the scope of what one engineer could manage, and companies

22:38always expanded to feel that capacity.

22:41Why?

22:42Because their competitors were expanding as well.

22:45If you can deploy faster, you build more features.

22:47If you can manage more infrastructure, you launch in more regions.

22:51The demand did not shrink.

22:53It grew with every such advancement.

22:57And with ai, the pattern is the same.

23:00One engineer will manage 50,000 containers or a hundred thousand serverless functions

23:05or complex multi-cloud deployments.

23:09That would have been impossible before.

23:11But this is important to understand.

23:12Companies will not say, great.

23:14Now we need fewer engineers because we are gonna keep managing

23:18the same number of servers and containers and functions as before.

23:22They'll say, great.

23:24Now we can build that new AI product that we have been planning.

23:28Now we can expand to these new markets.

23:31Now we can build more products and features.

23:33Now we can deploy in additional regions.

23:36Much faster with more scale.

23:38So the ticket queue never goes to zero.

23:41There is always more work or more to do, and AI makes an individual

23:46engineer faster and more efficient, but it doesn't make you obsolete.

23:51Now let me show you one of the most practical use cases where AI

23:55is already providing value today.

23:58And this is something that you can implement right now.

24:01I think every observability monitoring tool's gonna have, if it doesn't already,

24:05it's gonna have AI features to help you troubleshoot faster, get to the problem

24:10resolution faster, make suggestions.

24:12We, we've seen some of these.

24:13Post, they will see an issue that's triggered and in alert systems.

24:17So whether you have PagerDuty or whatever the PagerDuty notify the ai, it accesses

24:22the metrics and logs and data that guesses on what the uh, resolution might be.

24:26And then it posts that into Slack, right?

24:28So then like you as the human, when you're woken up in the middle

24:31of the night, you automatically.

24:32Look the slack at the issue and right below it is an AI response going.

24:35I've looked at the data, it might be this, this, or this.

24:38Like, I think that's the very first thing that ops people should be looking

24:41at because that, you know, reducing your time to fix is a key DevOps

24:45metric that we've all been working on.

24:48And AI, I think can give us that first pass at least to narrow down, you know,

24:52hey, I looked in the pod spec logs for that pod that was alerting and it's,

24:55it's, I see these three errors and I filtered it down for, those are the

24:58things I think that are gonna happen.

24:59Yeah, I think this specific use case that you just mentioned about

25:02AI layered on top of observability because you have tons of metrics,

25:06you have tons of data, so it kind of.

25:08Is a perfect use case for AI and analytics and then coming up with a suggested

25:13solution because that's probably one of the things that humans do not like to

25:17do or not enjoy doing, and especially reducing the time to fix the issue

25:22when it's actually very critical.

25:24I've heard that actually.

25:25Is one of the top use cases whenever, you know, it was a

25:29discussion about AI in the context of

25:32DevOps.

25:32And it can be read only.

25:33Like a benefit of that is it's read only.

25:35You don't necessarily give need to give it right to your infrastructure.

25:37So it's a low risk, like if a guess is wrong.

25:40You didn't waste your time, right?

25:42You just have to troubleshoot more.

25:43This is one of the most practical AI use cases for DevOps right now,

25:48alerting and initial troubleshooting.

25:51So here is how it works.

25:53Your monitoring system fires and alert AI automatically pulls all the

25:58relevant logs, metrics, traces, and any recent configuration changes made.

26:04To the environment.

26:05AI then analyzes all this data and suggests possible causes.

26:11What caused this issue?

26:13Again, AI posts its analysis from all this aggregated data to Slack

26:20channel alongside the alert, describing exactly what happened and why.

26:25When you wake up at 3:00 AM you immediately see exactly what the cause of

26:31the issue was and the potential solutions.

26:34This saves the first 30 to 60 minutes of troubleshooting that time when

26:40you're finding which pod is failing, pulling all the logs, checking

26:44any recent deployments, any recent changes, looking at the metrics,

26:48searching for similar past incidents.

26:51AI can do all that automatically and much faster, and it will just present

26:56you with, here are the three most likely causes based on the data, but here

27:01is why this use case works so well.

27:03Because it's read only, the AI is not changing anything.

27:07It is just analyzing and suggesting, so if it's wrong, you lost a few minutes.

27:13If it's right, you just saved 30 minutes of troubleshooting.

27:17But it doesn't actually have a risk of messing up anything in production

27:21because it doesn't make any changes.

27:23So low risk, high value.

27:26This is exactly the pattern you want for AI in operations.

27:37One theme kept coming up in my conversation with Brett.

27:41Context.

27:42AI needs massive amounts of context in order to work well, and the biggest

27:48misconception that people have is that.

27:50AI will just magically understand what you want with a single prompt.

27:55So let me show you what actually happens in reality.

27:59So you said this phrase of AI is not gonna do your stuff, so you still

28:04have to babysit it, whether it's doing DevOps related task or code generation.

28:08And a lot of people, I think the biggest misconception they, that

28:12I see that people have mm-hmm.

28:13When they do not know.

28:15Have not done proper research of AI and its capabilities is that it's

28:20automatically is gonna, you know, you just give it one prompt and it's

28:23gonna magically do stuff for you.

28:25And when you start digging deeper and understanding how it works,

28:30what you said exactly like it needs a lot of instructions, it needs

28:33a lot of guidance, documentation.

28:35So once you get all of that.

28:38Knowledge down back to the original story of you realize the mistakes, so you start

28:42giving it more context and more context.

28:44At the end of it, you'll realize that your ai, maybe it's weeks or months later, but

28:47your AI is actually way more reliable now.

28:50It can take those DevOps tickets from your GitHub issues or your Jira or whatever.

28:56It can take those tickets and then it can get like 90% of the time it gets it right.

29:01The first time maybe, and you're maybe getting pretty close to almost

29:04trusting that it, that it can do a job.

29:06Yeah, but it's, what it's really doing is it's doing a very narrow scope.

29:10The documents are very well written.

29:11The requirements for the poll request or the change request are very specific.

29:16And because it can feed from all those different places.

29:20It suddenly has, you know, it's probably using the hundreds

29:23of thousands of tokens now.

29:24It might even have access to your metrics to understand.

29:29The biggest misconception about AI is that you give it just one

29:33prompt and it does everything.

29:35But in reality, ai.

29:37Needs extensive instructions, documentation, context,

29:41data, and guidance.

29:43Think about when you onboard a new engineer in your team.

29:47You're not gonna say, just set up our infrastructure and walk away.

29:50You give them architecture documentation, you explain your

29:54naming conventions, you tell them what the security requirements are.

29:57You give code examples, access to the existing systems.

30:01And you also tell them, this is a person that you can go with your

30:04questions if you need any help.

30:05And AI is the same except it cannot ask for clarifying questions.

30:10So you need to provide even more context upfront and notice the timeline

30:16that Brett mentioned weeks or months.

30:18So this is not instant.

30:20You don't get AI working perfectly on day one or day two.

30:23You start with 70% accuracy.

30:26Then you give it more context, you document better, you refine your prompts.

30:31You narrow your scope and gradually over weeks or sometimes a month, you might

30:36get to 90% accuracy for specific narrow tasks, but that 90% accuracy only happens

30:44when your documentation is excellent.

30:47Your requirements are specific.

30:49The task scope is narrow and you have fed hundreds of thousands of tokens of

30:54context, and this is why companies with mature DevOps practices will benefit

30:58more from AI because if you already have infrastructures code, properly organized,

31:03CICD pipelines, well documented, run books for common tasks or architecture diagrams

31:09and clear standards and conventions.

31:12Then AI can learn from all of that.

31:15But if your infrastructure is undocumented, if you're doing a lot

31:18of the stuff manually, your standards are inconsistent and everything is

31:23just tribal knowledge in people's heads, AI will not help much.

31:27Because it has nothing to learn from.

31:29So the work you do now to organize and document your systems, that's

31:35an investment you are making to use those AI tools effectively,

31:39or maybe your actual live Kubernetes endpoints.

31:42And it might reach out to see what the real world is looking like right now.

31:47And that's something we're seeing is where if you tie in these MCP tools,

31:50which MCP basically just gives your AI access to all your other APIs.

31:55And so it might have access to all of your remote cloud tools.

31:59It has access to GitHub through the API.

32:01It has access to AWS through the API, so it can start to read

32:04and look at things all the time.

32:06You're slowly refining it and getting the scope down, and

32:09right now it's loosey goosey.

32:11Everything goes.

32:12Everyone's just.

32:12Throwing everything into it.

32:14It's kinda like the early days of containers where we were throwing

32:16everything in our containers, including all the bad stuff.

32:18I think it's a maturity model and you're, you're all gonna get there,

32:20but you have to start now or, or wait, but start now, and then maybe

32:24there eventually you'll, you'll have something mature in six months.

32:27MCP or model context protocol is becoming really important for this reason.

32:33Right now, when you use ai, you are basically just copy

32:36pasting information, right?

32:38You copy your Terraform code into chat GP team.

32:41You copy error logs into cloud or use it in your IDE, so

32:45you kind of manually feed it.

32:48The context MCP changes that because it lets AI directly

32:53access your systems through APIs.

32:56It can read your Kubernetes cluster state, it can check your GitHub

33:00repositories and the code inside.

33:02It can query your monitoring metrics or look at your cloud resources and

33:07existing infrastructure, and it can access your documentation instead of

33:10you manually providing context and explaining how your infrastructure

33:15looks like or what your code is doing.

33:17AI can automatically pull the current state of your systems.

33:22It can pull the current state of the code.

33:24It can see what's actually running in production right now.

33:28And this makes AI much more useful for troubleshooting operations work.

33:34For example, when an alert fires AI can immediately check the logs of COR

33:40pods or look at recent deployments or query metrics from the last hour.

33:45Review any configuration changes.

33:47And based on all these rich data, it can suggest any potential causes.

33:52But Brett's warning is important.

33:54Right now.

33:55Everyone is just throwing everything at ai.

33:58No structure, no security model, no guardrails.

34:01This is like the early days of containers.

34:04People were putting literally everything in containers, including passwords

34:08and secrets, all the bad practices.

34:10And then we learned from all those mistakes and we matured.

34:14We learned the best practices.

34:16And AI is following the same path.

34:19We start now, we experiment, we make a bunch of mistakes.

34:22We learn and iterate, and then in six months you'll have something mature.

34:33Now, here's the part that surprised me the most.

34:36Brett spent two years walking around conferences.

34:39Asking one simple question and the answers revealed the actual

34:43truth about AI adoption in DevOps

34:46by our cloud, right?

34:47Because we're the ai, ai, ai, that was where this all came from, was like how,

34:51I don't even know, like I'm overwhelmed.

34:53I hear all this wonderful hype about all this code being written

34:55in ai, and then I go talk to my friends and they're like, Nope.

34:58That's not happening.

34:58Like I don't see it, like no one was seeing it.

35:01So I started asking questions and I started going around Kub Con.

35:04And for two years now, I walk around KubeCon.

35:07I live in the expo hall.

35:08That's all where we hang out the whole week.

35:09'cause I could watch the videos online.

35:10I don't need to go to the sessions.

35:11I, I'm there for people.

35:12And all week long I'm asking, Hey, do you run your own AI inference?

35:16Do you use AI in DevOps for everyone?

35:19For years, all the talks were telling us about it, but then

35:22everyone I talked to is like, no.

35:23No, no, not using ai.

35:25Not running ai.

35:26I'm using Chad GPT and I'm code jenning.

35:28But I'm not automating with AI and I'm not certainly touching infrastructure

35:32with ai, and I'm certainly not running my own AI inference cluster.

35:36So it was a weird dystopia, not dystopia.

35:38It was a weird disconnect between what I was hearing on the internet

35:41and what even some AI conference talks were talking about.

35:45And then what everyone I knew and everyone I would talk to, I was like

35:48a man on the street trying to find that one person that was using it.

35:51It just wasn't happening, and it wasn't until this year.

35:55That we finally started having actual DevOps ai conversations.

35:59If you'd ask someone a year ago, they would've said, that's crazy.

36:01I, I would never let an AI touch my infrastructure.

36:04But now we're actually starting to have conversations, so I feel like

36:07hopefully, hopefully I nailed the timing.

36:10Like not too early, not too late.

36:11'cause you can be too early.

36:12That's also another problem.

36:14This is such an important observation.

36:16For two years, Brett walked around CubeCon, the biggest Kubernetes and

36:20cloud native conference and he asked people, are you using AI endeavor?

36:24Conference talks said yes.

36:26Marketing materials said yes.

36:28LinkedIn posts said yes, but actual engineers, the practitioners who were

36:34supposed to be using those tools said no.

36:36They said, we are using Chet g BT to write code faster, but we're not

36:41automating infrastructure with ai.

36:43We are not letting AI touch production environments.

36:46And that means there was a complete disconnect between

36:49the hype of AI and the reality.

36:52Now, why does this matter for you?

36:54Because you need to separate what's possible in demos.

36:57What's.

36:58Being solved by the AI companies and what's actually being used in

37:03production by real engineering teams.

37:05The AI companies will tell you their tools can automate everything.

37:09The conference talks will also show these impressive demos of their AI tools, but

37:14those demos are often heavily scripted in controlled environments, perfectly

37:19prepared with perfect documentation and context and unlimited time to

37:24prepare in real production environments.

37:26You have messy systems.

37:28The documentation is in complete requirements change constantly.

37:32So the adoption is happening much slower than the hype suggests, and that's okay.

37:39That's absolutely normal, and it kind of fits into the pattern

37:42of new technology adoption.

37:44This is more of a slow step-by-step adoption.

37:47And Brett says it's only this year, so 2020 4, 25.

37:53That real conversations about AI in DevOps are starting, not implementation

37:58at scale, just conversations about how to implement it safely.

38:03So that means we are still in the very early days.

38:06You are not behind.

38:07You are actually right on time.

38:10Now, I wanted to understand what Brett's vision is for the future.

38:14What does he think AI in DevOps will actually look like?

38:18When it matures and his answer paints a really interesting picture.

38:23I don't

38:23want to ever be called out of like, you're teaching this, but you have

38:25no idea what you're talking about.

38:26Right?

38:27Like, I really hope that I don't, I'm sure it happens, but I really try to

38:29hope so I'm really excited right now.

38:31What will my AI workflow in DevOps look like?

38:35Once I've taught everyone else all the steps, because I don't quite

38:39have that yet, and in fact the industry doesn't have that yet.

38:42I'm imagining this future six months from now where I have, uh,

38:46I have a lot of GitHub examples.

38:47For years I've had, I have DevOps and Kubernetes examples and all sorts of

38:52GitHub actions examples, because I've been teaching that stuff for a decade.

38:55I'm imagining taking all those repos or one of those repos and turning into

38:59like a fork that's an AI version where literally everything is done by the ai.

39:04Maybe this isn't like.

39:04Future, like actually gonna be how production is is done.

39:07But every step is done by ai.

39:09All I've gotta do is put in an issue.

39:11Everything else happens after that.

39:13And at the same time, it is way more thorough.

39:17So it, it does automatic security checks for CVEs, and then when it sees the

39:21CVEs, it actually recommends how I could fix those CVEs in my images or my code,

39:27or you know, whatever I have to deploy for somebody when I make a terraform.

39:30Commit it, it reviews it and gives me feedback on how to improve it.

39:34Like I'm, I'm imagining this scenario where I can actually,

39:38hopefully, eventually know less about what I'm doing and the AI

39:41provides me the safety guardrails all the way through my pipeline.

39:45I don't think it's any one tool specifically.

39:46It may not answer your question, but I think it's understanding.

39:50The end-to-end workflow of what the current ais can do for me after commit.

39:55'cause I kind of look at my job as like, I'm the person af the developer

39:58who's making the next, the next Facebook or whatever their job is

40:02to make the best app for users.

40:03My job is to make the best experience for them.

40:05Uh, that's how I look at it.

40:06Them and the users.

40:07The users are my, my users, but, but the developers themselves are my users.

40:10And what if I could, like, there's an old show called

40:13Silicon Valley, I still love it.

40:15You should.

40:15Absolutely.

40:15If you're not a fan of the show, if you've ever watched the show, Silicon

40:18Valley is absolutely a decade later.

40:20I think it just hit its 10 year anniversary.

40:22A fantastic show that is that, that actually talks about

40:25AI and they say inference.

40:26From like eight years ago that I didn't even know what that word meant.

40:29I watched that show again recently and it's absolutely still relevant

40:32today for like this current chaos of AI is perfect for today.

40:35They should just rerun it.

40:36I had this idea that like there's this character called Gilfoyle.

40:39He's basically the DevOps engineer and he is able to step away

40:43and replace himself with an ai.

40:45The AI chat bot's like chatting away with his colleagues saying,

40:47saying that he's, that it's doing the work and it does things.

40:50He's not working right.

40:51He's still gotta be there because he's gotta manage the

40:53ai, but he's not really working.

40:55He's just sitting back drinking a, a cocktail.

40:57I am kind of imagining is, are we there?

41:00Are we close?

41:01I mean, no one's really shown that yet.

41:02There is certainly no one at conferences talking about that yet.

41:05Even in London, six months ago, there was at all of KubeCon.

41:09Hundreds and hundreds of talks.

41:11One, there was one talk that detailed in depth their efforts

41:15of trying to reduce toil by having an AI review all their DevOps prs.

41:20And it turns out at least six months ago, it was way harder

41:23than I was interested in doing.

41:25Like I watched that it caused me to go, nope.

41:28Not for me, not, not yet because there was so much to it.

41:31And I'll, I'll, I'll up the link in the show notes.

41:32That to me is like, how can I mix and match?

41:35That's what I'm really interested in.

41:37I don't think there's any one tool DevOps is gonna use to magically solve

41:40all these problems, but I am very interested in the workflow pipeline

41:43aspect of what if I replace myself with an ai, how crazy would it be?

41:48How wrong would it be?

41:49And obviously we want safety and all that, but just experiment.

41:52See where, how far you can get, and then you sit back and you can

41:55drink, uh, your MI ties or whatever.

41:57This vision is fascinating because it shows both the

42:01promise and the current reality.

42:03Brad imagines a future where you open a GitHub issue and AI handles everything.

42:09After that, it runs security scans.

42:11It reviews your Terraform code, it checks for CVEs and suggests any fixes.

42:17It provides guardrails throughout your entire pipeline.

42:20Okay, but notice what he says.

42:22I can know less about what I'm doing, and the AI provides the safety

42:28guardrails, and this is a key shift here.

42:31The value is not in memorizing yml syntax anymore.

42:35The value is in understanding the workflow well enough to set up AI

42:41correctly, to give it the right context to review its output.

42:45To know when something looks wrong and needs correction, and the Silicon

42:51Valley reference is perfect here.

42:53The DevOps character in the series replaces himself with an

42:58AI chat bot, but he's still there.

43:00He still manages the AI behind the scenes.

43:03He just automates the repetitive work, and that is the realistic future.

43:08Not AI replaces DevOps engineers.

43:10But DevOps engineers use AI to handle the tedious parts.

43:15So they can focus on higher level problems, but here is the reality check.

43:20Even at CubeCon with hundreds of talks, there was one talk about

43:25actually implementing AI for DevOps prs, and it was way harder than most

43:32people want to deal with right now.

43:34So the vision.

43:36But the path is unclear.

43:38We don't know how fast we're gonna get there.

43:41The tools are still immature, but people are experimenting and that's

43:46how every new technology starts.

43:55Do you think that the value of engineers will switch more towards, um,

44:00understanding the context, understanding the use cases, the logic rather than

44:05memorizing the syntax of the tools or even specific tools and how they work?

44:10Because maybe that can be automated and, and done with ai where AI kind

44:15of does this tool selection maybe at a granular level, like do you think.

44:20That's where the value of engineers or the future requirements of engineering

44:25skills will shift to where they're more like architects and designers

44:28rather than people who actually.

44:30Do the execution or implementation, what do you think that evolution

44:33is gonna look like for the engineering skills in general?

44:36Again, trying to read the tea leaves, trying to predict the future

44:38is always tricky with this stuff.

44:39But if we can look sort of at the last two years of progression, we are

44:43nowhere near a GI for DevOps, right?

44:46We're nowhere near.

44:47An AI that you can just hand over the keys of the kingdom and it does the

44:51things without you just thinking about it.

44:53So in my mind, you still kind of need to know all these things.

44:57You do need to know the syntax because how will you know when it's

45:00wrong and the AI can't be fired?

45:03The company isn't going to fire Andro because it deployed a Kubernetes

45:08cluster with the API for Kubernetes, accidentally public on the internet.

45:12The person who committed the AI code is gonna get fired or at least in trouble.

45:16Right.

45:16Maybe in a world of the far future where we think of AI as having rights and jobs

45:22and paychecks, and it can be fireable and hireable, maybe in that world,

45:26but we don't treat them that way yet.

45:28I don't see any sign of that happening anytime soon.

45:29So there's still gotta be a human that has to go, I'm willing to take this risk.

45:33I'm willing to commit that button.

45:35Push that commit button or push that deploy button until

45:38that one step can be removed.

45:40I think we have to be able to know what it's doing.

45:42I think that every team manager, when things don't go well and production goes

45:46down, or we have a deployment that fails and rolls back, there's always gonna

45:49be that DevOps, engineer management that's gonna say, okay, let's do.

45:53A root cause analysis.

45:55Let's ask the the five why's and the people in the room are gonna

45:58say, well, the AI did this well, why the AI decided to do this?

46:01Well, why?

46:02This is probably the most important insight in this entire video.

46:07Lemme break it down to you.

46:08The question I asked was, will engineering shift to being more

46:12about architecture and design while AI handles the implementation?

46:17And Brett's answer is clear.

46:20Not yet.

46:21Not for many years.

46:22Why?

46:23Because someone needs to be accountable.

46:26When your Kubernetes cluster is accidentally exposed to the

46:30public internet, your company cannot fire Claude or Che GBT.

46:35They're gonna ask which engineer deployed this, which human approved this?

46:39Which person takes the responsibility and that human better understand

46:44what they deployed and why.

46:46You can't say AI did it in a postmortem because that won't save your job, and

46:50this is why you still need to know the syntax and the implementation details.

46:55You still need to understand how Kubernetes works on a low

46:58level, not just high level.

47:00You still need to know what good Terraform code looks like versus bad

47:03Terraform code, but not because you are writing it all from scratch, but

47:07because you need to evaluate and assess.

47:11What the AI generated, you need to spot the mistakes.

47:14You need to know when something is wrong.

47:16So the skill shift is actually subtle, but very important.

47:20Before ai, you write infrastructure code from memory or from documentation,

47:26copy pasting some code snippets.

47:27With ai, you get the output from ai.

47:31You review and verify infrastructure code.

47:34That AI generated, but both require deep knowledge.

47:38You can't review something you don't understand.

47:40You cannot spot errors in code you've never learned.

47:43So the fundamentals still matter.

47:46Learning Kubernetes, learning, Terraform Learning, CICD, in the syntax of

47:50all these configurations, learning, networking, and security, you need

47:54to understand not just high level of.

47:56How these things work, but also the syntax and configuration details

48:02and the basic foundational knowledge of all of these topics, and then

48:07you add AI on top as a tool to work faster with all of these tools.

48:13But don't skip the foundation thinking that AI will replace it.

48:17Because when things break at 3:00 AM you need to understand what's

48:21happening and how it needs to be fixed.

48:24The AI may help troubleshoot it, but you are the one who needs

48:28to fix it or be responsible for the final fix that gets applied.

48:40So after hearing all of this, the limitations, the possibilities, the gap

48:45between the hype and reality, the question becomes how do you actually start?

48:50What's the practical first step?

48:52And I asked Brett this specifically because it's easy to feel

48:57overwhelmed in this whole AI.

48:59Ecosystem right now.

49:00There are hundreds of AI tools, hundreds of workflows that you could automate.

49:06So where do you even begin?

49:08And I believe his answer is going to save you a lot of wasted time.

49:14So what's the practical strategy?

49:16How do you actually get started with AI in DevOps, given everything we've discussed?

49:21And so that to me is most people step one.

49:24I don't even think I even have to tell them that, that they're

49:26just all gonna do that first.

49:27But that doesn't necessarily correlate to AI in my CI in terms

49:32of how I'm gonna set all that up.

49:33It's gonna run somewhere else.

49:35I'm gonna pick a bunch of tools that already exist to start with.

49:38Maybe my team already uses cloud code for everything, so we're

49:40just gonna use cloud code and ci.

49:42If your team has experience with that, do that because you're gonna

49:46save yourself time of learning.

49:47The nuances of prompt engineering and how do I write prompts for

49:51an AI so it doesn't hallucinate.

49:52I gotta tell it.

49:53It's it's gotta do it.

49:54Great.

49:54Or we might have to hurt grandma.

49:56Like you gotta threaten it.

49:57Hopefully grandma will be fine.

49:58Hopefully we don't have to mess with grandma much longer.

50:00There's another category, which is not necessarily the category of what AI

50:04will do, but the buy model is quickly.

50:08Like everyone, right now, all these small startups are all in a

50:11race to be the DevOps AI company.

50:14I'm not necessarily an expert on any of them, but I have been paying

50:17attention to that market and how they're, they are all trying to.

50:21Basically shortcut like any other build versus buy model.

50:24Like I always tell people you're either gonna pay for it with

50:27money or you're gonna pay for it.

50:28With time and open source, you usually pay for it in time, and if you're

50:31gonna buy, you usually pay for it with real money and it saves you time.

50:34The same thing is happening with AI for DevOps, we're now slightly aware

50:37of maybe half a dozen companies that they're trying to be the one shot AI for

50:42DevOps automation tool, and you use their platform rather than putting together

50:46your own things through your own scripts.

50:48Turns out largely their problem.

50:50Is context.

50:52So a lot of them are spending a lot of their time training and specializing

50:55these models to be really good at DevOps.

50:58Tasking,

50:59here's the practical strategy.

51:00Step one, start with the AI tools that you are already using.

51:05If your team uses Jet GPT, use that.

51:08If you're using cloud in your team, use that.

51:10Don't try to learn five new tools.

51:13All at once.

51:14Step two, pick one specific narrow workflow to automate like CICD.

51:20Pipeline generation is a good start, for example, or writing Kubernetes

51:24manifest files or generating Terraform modules, something specific and isolated.

51:30Step three, document your require.

51:33Clearly, this is critical.

51:35The more context you give the ai, the better it performs.

51:39So write down your naming conventions, your security requirements, your

51:44standards, and feed that to ai.

51:47Step four, review everything.

51:49Do not trust the AI output blindly.

51:52Check it, test it, make sure it works, even if it works, that

51:56it follows the best practices.

51:58And so on.

51:58Step five, iterate.

52:00The first attempt will most probably not be perfect.

52:04You will give AI more context.

52:06You will refine your prompts.

52:08You will narrow the scope, and over weeks or month, it will get better.

52:12Now, there's also the buy versus build decision.

52:16You can build your own AI workflows with open source tools, or you can

52:20pay for a platform that specializes.

52:23In DevOps ai, for example, and built takes time, but gives you control

52:28by costs money, but saves time.

52:31So depending on what's your priority, you are gonna need to

52:35trade one thing for the other.

52:37The key is start somewhere.

52:40Do not wait for the perfect tool or the perfect workflow, or for AI to mature

52:44enough, but also don't panic and try to.

52:47Bring tens of AI tools all at once in your projects with the fear to not stay behind.

52:54You have time, so pick something small and start experimenting.

52:59That's how you learn and that's how you'll be ready when the

53:03AI tools actually mature.

53:06So after two hours with Brett.

53:08Here is what changed my perspective completely.

53:12Everyone is asking the wrong question.

53:14They're asking, will AI replace me?

53:17But the real question is, what happens when one DevOps engineer can manage

53:2250,000 containers instead of 1000?

53:25History gives us the answer here.

53:27Companies do not downs.

53:29They expand, they build more products, they enter new markets.

53:34They improve reliability.

53:35So the work never shrinks.

53:37It grows.

53:38Now, here's something that most people completely miss.

53:42AI does not eliminate the need for DevOps knowledge.

53:47It actually makes it even more important because think about it,

53:50when your monitoring system fires an alert at 3:00 AM and AI suggests.

53:56Three possible causes.

53:58How do you know which one is right?

54:00You need to understand your infrastructure deeply enough to

54:04evaluate all those suggestions.

54:06When AI generates a Terraform configuration code, how

54:09do you know it's secure?

54:11You need to know what good terraform looks like versus dangerous terraform.

54:16So the skill is not changing from doing to not doing, and

54:21just delegating it off to ai.

54:23The skill is changing from writing to evaluating, but evaluation requires

54:29even deeper knowledge than writing.

54:30You can't spot mistakes in code you don't understand, and this is

54:34why Brett's advice is so practical.

54:37Start small, pick one workflow, maybe CICD, pipeline generation or monitoring.

54:43Give AI clear documentation about your standards.

54:47Let it generate the pipeline.

54:49You review it.

54:50You learn what it gets right and what it gets wrong, and over weeks

54:53and months, you build that muscle.

54:56The muscle of working with ai.

54:59Not replacing yourself, but augmenting yourself.

55:02And the disconnect that Brett discovered at conferences is

55:06actually good news for you.

55:07It means you are not late.

55:09The hype says AI is everywhere and it's.

55:12Super magical, but the reality says that most engineers are

55:16still figuring out where to start.

55:18So start now, not because you will fall behind, but because engineers who

55:23experiment today will have month or years of experience when these tools

55:28actually mature and document your systems.

55:32Not anymore for humans, but for the AI that will help you work

55:36faster next year or in two years based on that documentation.

55:41And very, very importantly, learn the fundamentals, not

55:45despite ai, but because of ai.

55:49And final one is stay skeptical.

55:52When someone promises full automation with ai, ask them, who have you talked to?

55:58That's actually running this in production and getting the

56:02results that you are promising.

56:04Now, if you want to continue learning about this topic of AI for DevOps,

56:09make sure to check out Brett's podcast.

56:12It's called Gentech DevOps.

56:14He's interviewing people who are actually implementing this stuff

56:18in production, so not some theory and opinions, but real use cases.

56:23Which is how you should be learning everything.

56:26And let me know in the comments, are you already using AI in your DevOps work?

56:31And if yes, what workflows have you tried to automate already?

56:35And share in your comment what worked for you and what didn't.

56:39I wanna hear your real experiences.

56:42And people who have not tried AI can see the realistic

56:46picture of AI usage in projects.

56:50So live your insights and experience down below for the whole community.

56:54And with that, thanks for watching and I'll see you in the next video.