DevOps, AI & the Future of Software Engineering

00:00:00: teams that have already adopted the agile principles.

00:00:03: They are getting faster and they're getting faster at everything, so AI and Agile using each other to accelerate.

00:00:12: it's quite massive.

00:00:14: when you have everything in place how much acceleration You have?

00:00:17: The teams don't have that.

00:00:21: as I said there is a big trouble.

00:00:25: Welcome to Hello!

00:00:27: The podcast where we break tech bubbles before and where we talk to people who make tech more humane.

00:00:59: He has been active for many years in software engineering DevOps, leadership and community building.

00:01:04: And he's also closely involved in the DevOps Community in Switzerland through initiatives like DevOps Days.

00:01:10: In this episode we talk obviously about DevOps and what happens when a software team is optimized for shipping features.

00:01:17: We explore how processes roles and the craft of Software Engineering are changing right now because of AI.

00:01:23: Let's dive-in!

00:01:30: Welcome Romano to The Hello!

00:01:33: Thank you very much for having me.

00:01:34: So two weeks ago there was the DevOps days, this conference that your core organizing since many years?

00:01:40: How is it this year for you?

00:01:42: It's absolutely awesome!

00:01:43: It was wonderful and what I really liked, it's again that we were fully booked.

00:01:55: So two hundred and fifty people over there And it was very energizing.

00:02:00: a broad flower of topics We discussed.

00:02:04: they're as it was really cool!

00:02:06: What are some the main takeaways for you this year?

00:02:08: What is like topic stood out?

00:02:10: Of course AI This time when my menu sort-of look back at the open spaces.

00:02:18: So we also do a little bit of unconference there, then you could see that AI was the dominating topic not anymore like GitLab versus Jenkins all off the tooling stuff.

00:02:30: it was more how Do You Do DevOps with AI?

00:02:33: and does DevOps survive?

00:02:36: With AI?

00:02:38: these were the main topics.

00:02:40: Anything that surprised in their conversation.

00:02:42: what surprised me is a little sort of getting irrelevant as an engineer.

00:02:50: That is something that you always hear from the people, but for me yes on one side it's surprising and also see all tech CEOs are advertising exactly that topic, and they are doing whenever they can sort of a vendor pitch.

00:03:13: That AI will remove all off the jobs?

00:03:16: We'll definitely get back to this question later on.

00:03:19: I'd also be curious.

00:03:19: so i've never been at the DevOps days.

00:03:21: And who were the people that are attending this conference?

00:03:24: Why Are They There?

00:03:26: So When you look at the DevOps days, of course there are many engineers.

00:03:32: But also leadership persons and UX people are here as well all of the system engineers joining us And what we do now can see is that more and more people from the QA community attending or business analyst community.

00:03:51: So

00:03:52: you said it's the ninth time that it happened and you've been involved for many years as well.

00:03:57: Have you seen this conference evolve also in terms of people attending, and topics discussed over the last couple of years?

00:04:03: When we look at topic stuff then first was more about tooling.

00:04:08: what kind of CI CD pipeline?

00:04:10: how does a CI CD Pipeline looks like?

00:04:13: What tools are is best tool?

00:04:15: Then of course Kubernetes the container station.

00:04:19: topic came up sort of the theme until, um, two thousand and eighteen roughly.

00:04:26: Then that whole culture stuff came also quite up.

00:04:31: it moved then into their direction.

00:04:32: off platform engineering And from platform engineering we are now going in to the direction of AI.

00:04:39: when It comes to there too The people what you could first could see is?

00:04:44: That at the beginning it was engineering heavy uh...and Also We had quite a long number of female people attending.

00:04:53: We changed then that we also introduced our own diversity concept, and now we have roughly a number of fifteen percent female attendees.

00:05:05: My opinion still too low but unfortunately there are not more people in the university.

00:05:12: so it is a low number But its higher than the people coming from University.

00:05:19: It's super interesting, a diversity concept for our conference.

00:05:21: Can you share a little bit more about that?

00:05:23: Yeah sure!

00:05:24: So we have the diversity concept but we also have a concept of sustainability.

00:05:30: it comes from us because we believe that diversity also fuels quite good discussions and also culture And this is why we introduce the diversity concepts.

00:05:43: We make them open to everyone And I mean, really for everyone.

00:05:49: So this year at the first time we had a person that is Dief At The Conference and he joined us with two sign language translator.

00:06:01: Yeah!

00:06:02: They were there and they translated the whole text to him.

00:06:06: That was really awesome.

00:06:08: We are very open and we want to encourage you so much.

00:06:12: Everyone can join when you have more women Because we are a conference that have talks, we have ignites and then you can see in the discussion is different.

00:06:29: You talk about different topics not only technology but also culture leadership or soft topic.

00:06:39: For me, this is the important thing because when you look at organizations then organization are socio-technical systems and in a social technical system of course here.

00:06:51: You can fix stuff by adding an if statement into code but that's not big thing to do.

00:06:58: In most cases for example need to fix the organisation processes leadership or how we're talking with each other.

00:07:08: These are things where when you have a diverse group of people, this is much better doable.

00:07:15: Let's now shift gears and talk a little bit more about DevOps in itself because you've course also know that DevOps often competing against other types of work we do in software development such as shipping features or delivering something on time to customer who has direct intangible business value.

00:07:31: DevOps is often seen sort-of the afterthought for some thing it just done at side.

00:07:36: how?

00:07:36: Do you see the values of DevOps also changing?

00:07:39: maybe how people appreciate it over the last decade.

00:07:42: When you look back in, there was that awesome book called Accelerate from Forescreen.

00:07:49: what she did is she analyzed on a scientific level What DevOps does to companies or value streams and product teams?

00:07:57: And when she found out And then you are faster, have a better quality.

00:08:06: You have the fastest time to market more value for money all of these things and she looked at it from scientific perspective also.

00:08:13: right now there is regular Dora report that's coming out every year which also proves when you apply DevOps in company... ...you're faster and can create more values.

00:08:29: Yes, I also see it sometimes that reluctance especially from management.

00:08:34: The thing is... It's not about DevOps as this sort of a natural thing.

00:08:39: you have sort-of working system and this working system needs to be fixed.

00:08:44: You need to change something And what is behind all of that?

00:08:49: Is just a change resistance.

00:08:51: And of course, change always hurts and it also sometimes means that you need to reorganise.

00:09:00: Which is even more painful in organisations.

00:09:04: but nevertheless I say what would you like?

00:09:09: Do you want to stay where are or do you want sort of generate more value?

00:09:16: And in the end, many companies are going in direction of generating more value.

00:09:22: And as a consultant at Tilke, you've probably seen different types of companies and organizations trying to apply DevOps.

00:09:30: How do see this like maybe small company or small organization?

00:09:34: how they are tasked with introducing that versus big organization?

00:09:39: In smaller companies it's quite natural.

00:09:42: because you have smaller teams everybody is doing everything which just means you are organized along the value stream.

00:09:50: You don't have sort of any silos in there or any bottlenecks and handoffs, so value flows clearly in a small company.

00:10:00: So usually in small companies DevOps is sort of the standard even if they don't call it DevOps.

00:10:07: When you look at large companies, then usually you have a functional organization where they are organized in for example IT and business operations And usually the value does not flow through these silos.

00:10:24: So what you end up is with some handovers over the wall of confusion, how I call it?

00:10:31: Usually and The problem Is that the value Does Not Flow Through That System When You Have These Big Companies Usually By Applying For Example Value Stream Mapping You Can Identify All Of These bottlenecks.

00:10:46: But of course, it always comes down to we need to change the process or we needed to change your organization.

00:10:54: And this is usually something that bigger corporations don't want you do?

00:10:59: You mentioned value stream mapping.

00:11:01: I would say there's probably a bit of consulting term.

00:11:04: Can maybe use an example Or like concrete story That you've seen yourself how these could turn out and what was the outcome Of such a mapping process?

00:11:14: So the values remapping is nothing absolutely new.

00:11:16: It's quite an old term, again it comes from the Toyota production system.

00:11:22: The intention behind that is to apply whole systems thinking What you want a clear view on how value is produced in product or service.

00:11:36: You are going to map every step that you have from IDEA until it is in production and generates a value.

00:11:44: All of these steps, this sort the core of Value Stream Mapping Of course also identify persons And are going to measure how fast value flows through system and quality.

00:11:59: Based on that Value Stream mapping You can easily identify just visually on that where the bottlenecks are.

00:12:08: And usually such a value stream mapping is conducted in one day workshop, so you identify to current values stream together with put everyone together that is working from idea until production in to a room.

00:12:24: You have a bunch of post-its and then you go just, what are the steps for ideation at health production?

00:12:29: Go through it identify people or also identify your measure value stream And based on this you envision future value streams.

00:12:38: where do we want them when they remove all these bottlenecks.

00:12:44: And based on that, you then can derive action items for example adjusting the processes, adjusting technology and perhaps also adjusting the

00:12:57: organization.

00:12:58: go through this process.

00:13:00: The funny thing is when I do value stream mapping, i usually bring in also the management and management usually calls me and says hey uh our development team they are not fast enough.

00:13:12: we conduct their values tree mapping.

00:13:14: something very interesting happens because we are going to measure how value flows and where we have a waiting time And the funniest things you can see that Well, value flows quite fast when it comes to the engineers and they are producing the code.

00:13:31: Of course you always have some things that can improve there.

00:13:35: but usually we see part at start where you lose quite a lot of time.

00:13:41: Usually it's one third to two-third off the time is lost from someone has an idea over too budgeting until we have the prioritization, until we half all of the requirements together.

00:13:55: that takes quiet A LOT OF TIME!

00:13:57: And this usually very surprising for people to see.

00:14:01: that because many managers are talking nowadays about Time To Value, Time To Market and seeing sort of the start or they are in charge, is a very good thing and usually we can also resolve that quite quickly.

00:14:21: Super interesting!

00:14:23: the bottlenecks are actually more before engineering starts.

00:14:27: DevOps is often, at least for me.

00:14:29: in my mind it's more.

00:14:30: rather once engineering is done and we want to ship something of quality where you see a lever also having a lever on front like when we start working?

00:14:43: Absolutely!

00:14:43: It always.

00:14:44: how do you see DevOps?

00:14:46: When you look at general meaning of DevOps usually say okay DevOps development and operation.

00:14:54: And of course, now in the meantime many people have also come up with other names like business dev ops which is Business Development and Operation or DevSecOps which is developed security and operation.

00:15:07: you can see it In The End.

00:15:08: what the core Is.

00:15:10: It's About Value Flow and the value flow does not start With Implementation where the idea is.

00:15:18: And this is how we apply whole systems thinking.

00:15:21: and it starts there, but of course when you talk about quality We are doing quite a lot of shift left nowadays so bringing all off the quality assurance right at beginning end.

00:15:34: When I mean with shift let's say don't me only to quality assurance bring that to code.

00:15:41: You usually also bring back to requirements engineering and already into the ideation and same applies also to security.

00:15:49: So you touched on quality, security.

00:15:51: there's other non-functional requirements that DevOps is basically trying to tackle right?

00:15:57: And before I said at the DevOps days it was a lot about culture.

00:16:01: so how do you see teams actually adapting such a culture of quality or other non functional requirements?

00:16:09: what are some things that works?

00:16:11: What actually work Is that you have a shared ownership on the quality topic.

00:16:19: what does not work is when you say I have a team.

00:16:24: That's sort of balls, quality or on top of it really need to have that share mindset for quality which means everybody is responsible for the quality.

00:16:36: And the funniest thing, now also with AI you can see that getting more and more important.

00:16:44: in the end it all boils down to good software engineering or engineering practices.

00:16:49: In the End You need to understand what do want.

00:16:54: So the specification of what you want to do is very important also nowadays with AI.

00:17:01: And then, uh... What you need to derive out that how are going to verify?

00:17:08: That's what we're doing too.

00:17:09: does it according to their specifications?

00:17:12: and this is sort off the second thing.

00:17:15: by doing that at beginning not at end You already shifting left building directly quality in.

00:17:25: So you mentioned AI, obviously we have to talk about this also because of your new role.

00:17:29: congratulations!

00:17:30: I see AI is changing the software engineering process in itself quite fundamentally.

00:17:37: and of course i want to also hear from you... The agile development as we know it?

00:17:42: Is it going to stay or is it gonna completely vanish because of AI?

00:17:48: A very good question.

00:17:50: so When we look at Agile, then Agile in its core is all about that you can react on feedback.

00:18:01: So it's a very close feedback loop now.

00:18:05: In the past these feedback loops were quite long and with Agile We said okay?

00:18:11: We introduced two week sprints And then we had to weak feedback loop.

00:18:15: Now this AI we have much faster feedback loops.

00:18:20: Now back to your question, do you need agile?

00:18:23: Of course we need agile!

00:18:24: The only thing is that the feedback loops are getting faster and faster but we need all of the principles and practices for the Agile movement.

00:18:34: And in the end when you look at software engineering... ...the practices –and thats quite interesting–are staying sort-of same as good practices.

00:18:46: Of course, nowadays everybody can vibe code their application and you are getting a result with vibe coding but it is not something that is maintainable.

00:18:58: If you want to professionally do product engineering or create a product then you need to apply software engineering practices right from the beginning.

00:19:10: And I always say AI best thing that could happen to us, because now you need to apply all of the DevOps principle and all off the chat principles.

00:19:23: I've also heard recently that teams that already have adopted such a culture of quality and they already have automated testing implemented then they already do this.

00:19:32: it's much easier for them too start adopting AI or what we call agentic engineering right?

00:19:37: You probably also hurt that term.

00:19:39: And how do you see teams?

00:19:44: Yeah, they are in big trouble.

00:19:47: There also some reasoned paper out there which looked at it.

00:19:51: so as you just said teams that have already adopted the HL principles.

00:19:56: They're getting faster and they are getting faster had everything they are doing.

00:20:01: So AI and a child?

00:20:03: They are using each other to accelerate.

00:20:05: It's quite massive.

00:20:07: when you have everything in place how much acceleration You Have?

00:20:10: The Teams That don't have that, as I said they are in big trouble because now it is a negative acceleration.

00:20:19: They usually have bad quality code or they have bad tests and In the end what they are ending up Is just piling more bad things up Because AI Just looks at What's already there And It just generates again More Bad Code bad stuff in there.

00:20:41: And this is quite a bad thing for them, so the only things they are left of is adopting the HR principles and cleaning up all of the mess that they

00:20:53: have.".

00:20:53: In that sense AI also exposes these... problems in the teams, right?

00:20:59: And I think that's also a good opportunity to revisit some of these principles and really try to implement them before jumping onto the AI hype train.

00:21:09: basically.

00:21:10: Absolutely!

00:21:11: Can only second that you need to adopt these HL principles very fast.

00:21:17: if you don't do it You will have hard time Of course... ...you'll generate stuff quite fast but it will break and not be maintainable.

00:21:29: I also want to get back to this acceleration topic with you because at the beginning we said DevOps is a lot about people, what do you think happens in software engineering these days?

00:21:45: Do you see some negative consequences of that?

00:21:48: Oh definitely!

00:21:50: I also see it with me, these negative consequences.

00:21:54: I usually have six to eight cloud code windows open and they are all sort of working for me this AI agents And what i see is that quite a lot task switching tired thing.

00:22:11: So I get quite tired about working with that and i see it also with my colleagues.

00:22:17: the speed is just absolutely amazing.

00:22:19: but what you can also see, yes You have a lot of new technology coming out And people are having hard time to keep up.

00:22:28: What I already saw was some people getting symptoms of burnout influencers, they are also getting burned out.

00:22:43: So the speed is really a big problem?

00:22:47: Yeah I definitely want to stay for a moment on this because personally that's very important aspect of what we're seeing now with AI.

00:22:54: it is cognitive overload.

00:22:56: i would call there so much new stuff going.

00:23:01: love that they loved the input and always knew.

00:23:03: And then there's people that completely shut it off because they can't deal with it, how can we find sort of like a healthy way to participate in this change but also not as you said get burned out?

00:23:17: Now It all comes down to principles and practices In The End What We Need To Do As A Company?

00:23:25: We need Come Up With Some Principles that sort of stay as foundational principles and also some foundational practices.

00:23:34: And by doing that, we are creating an environment where everybody knows okay what stays.

00:23:43: because at the moment The problem is there's so much going on That many people don't know What really stays?

00:23:52: They don't have a common ground no foundation.

00:23:56: This is the tiring part, and what I usually do is sort of define what stays.

00:24:03: And what are things that change?

00:24:08: The second thing to do is invite people into a one day or half-day workshop where we are going to experiment with the new ways of working.

00:24:20: We're creating together, and giving people space that they can experiment because the worst thing you could do is just give them a license saying now use it!

00:24:37: because the problem is people have enough to do, but applying something new is unlearning and relearning of some thing.

00:24:47: And unlearning or relearning needs space for you need give that space.

00:25:02: there are two access on it.

00:25:05: does process stays or do we need to change the process at all?

00:25:11: And how much do you give to AI, and does it work together with humans.

00:25:20: When you think in these axes and define what works are done without AI then they get clearer for people but not under pressure.

00:25:35: Fully agreed, thank you so much for sharing that.

00:25:37: Staying on the people in software engineering how do you see the roles evolving?

00:25:41: Are we going to still have software engineers and QA engineers and designers In a couple of years or do you also see That some of these roles and tasks they're merging together more?

00:25:52: This is something quite interesting.

00:25:55: We also see that at the moment, roles are getting broader so they're enlarging each other.

00:26:03: for example The product owner or business analyst.

00:26:11: They can code small features already.

00:26:14: Also, when you look at the software engineer.

00:26:16: The software engineer also now creates architecture and of course it extends to quality assurance.

00:26:24: so they are getting bigger as well even if we look at Uwigs there too starting to code or do things from the product over nor from a business analyst.

00:26:36: So what I believe how the future will look like is that product teams get smaller In the agile world, we always talked about the team size of seven plus minus two people.

00:26:50: My prediction is this will collapse to three plus-minus two people and Of course with AI agents And I believe in the future everyone one of us will have four more agents that are supporting Us With The Work And this is also why I say we'll only have three plus minus two people in these product teams.

00:27:17: So, what kind of roles are those three people doing?

00:27:22: When you look at it... Of course!

00:27:23: You need to have a Product Owner and that can be supported by one team.

00:27:31: by four agents.

00:27:33: Of course, it can also be that for example he or she has someone with her which is very capable of designing stuff and then perhaps an engineer.

00:27:45: what I usually see you have different risks when you are doing product engineering And you want to leverage at that risk?

00:27:53: Of course in the future When You Have a Smaller Product these risks.

00:28:01: But when the product is bigger, then you of course also will have other persons.

00:28:06: So one of the risks are usability risk and usually we need to have an expert in usability.

00:28:13: Then another risk is feasibility risk which is engineering risk.

00:28:19: The third risk is financial risk.

00:28:22: so business case.

00:28:23: for that you need a product owner.

00:28:26: We talked already about skills or what's needed for people that work now in tech, and you mentioned also the conference.

00:28:34: That there's this fear of becoming irrelevant over the next couple years because of AI?

00:28:39: What sort of skills should people focus on these days to stay

00:28:42: relevant?".

00:28:43: In my opinion... The whole system thinking is getting more important Because one thing that AI quite bad at Is having a whole context and thinking in the whole system, looking at the whole systems.

00:29:02: Analyzing that whole system.

00:29:04: Optimizing this whole system.

00:29:05: This is getting more important.

00:29:08: The second skill is critical thinking.

00:29:11: In my opinion it's a very important skill for the future Looking at data, what AI is generating And having that critical mindset to adjust the AI.

00:29:24: That's an important skill.

00:29:27: Do you also see some new roles evolving.

00:29:30: So you mentioned the product owner will still be very important?

00:29:33: Are you also seeing like, new terms floating around that are now a bit more common in software

00:29:40: development?".

00:29:41: Yeah yeah I mean of course uh... You'll see some adaptations off-the-term and oh!

00:29:49: Of course there will also be new roles appearing.

00:29:54: but In the end, I think these roles that we already have they sort of stay.

00:30:01: The software engineer or system engineer will just get an AI label in any flavors at the beginning and also platform engineering but i'm pretty sure new roles will arise.

00:30:21: I asked the chief AI officer, this was a role that just recently has been generated.

00:30:30: We will see more of these.

00:30:33: Very cool and maybe to end this whole conversation how do you in your role as the Chief AI Officer?

00:30:42: What is something also gives hope for people working in DevOps or engineering in general?

00:30:52: I mean, when i look back at my career of twenty four years.

00:30:59: When the internet came everybody said now this changes everything and I had that thing over-the-years.

00:31:08: model driven development came a child came scrum came platform engineering and every time The people and especially marketing was telling me this changes everything.

00:31:21: Now we don't need these type of people anymore, We can now reduce the people.

00:31:27: It's sort a common theme that you see And also the fear Of new technology.

00:31:36: So when I come back to your question What gives my hope is it looks like That every thing changes But many things are staying always the same, and therefore I'm looking quite positive into the future.

00:32:00: Well now we're going to end with our rapid fire question so i'll have some questions for you or you can do quick answers?

00:32:07: Are you ready to jump in that?

00:32:09: Absolutely let's go!

00:32:10: Cool.

00:32:10: So first one if you could rewrite any line of code in any product that exists today what would you do ?

00:32:19: change in social media apps the default for having a public profile to private.

00:32:29: All right,

00:32:31: I think that would save quite a lot of challenges.

00:32:36: What's one everyday act of allyship?

00:32:38: That anyone can do

00:32:40: this is something that I do sometimes, but not too often.

00:32:46: In my opinion what you always should do is when have a meeting and see the person not engaged just sitting there or not discussing.

00:32:57: Just ask this person very gentle.

00:33:00: Of course I would like to hear your opinion on that, i really would love to hear you're opinion!

00:33:08: That's a positive thing and usually people in the room don't speak up but have good views of things going on.

00:33:19: Absolutely different perspectives always great to include.

00:33:23: The next one is the voices we think in tech.

00:33:28: In my opinion, we should hear more of the voices of women which are already eight to fifteen years in tech.

00:33:38: They're absolutely brilliant minds and they have an awesome view on the whole tech industry, on tech culture processes and organization And I would really love to hear more about their views.

00:33:55: feel very seen because I fall into this category and it's also a good sort of time frame to talk to these women.

00:34:02: Because that is the time where a lot of women drop out from tech industry for various reasons, so thanks for their shout-out!

00:34:08: And we'll end with last question – if listeners remember just one thing about this whole conversation what should be?

00:34:14: The important things are not technology alone but in the end AI is just technology.

00:34:24: Of course, it will have a big impact.

00:34:27: But what is important?

00:34:29: Is the organization that you are building?

00:34:32: so the processes of course to people and technology.

00:34:37: these three things already important thing.

00:34:40: in the end we as leaders We need create an environment where People can work with The Technology.

00:34:49: I think thats the important thing.

00:34:52: Awesome.

00:34:53: thank you so much for being here.

00:34:55: Thank you from having me.

00:35:03: what a timely and relevant conversation with Romano today.

00:35:06: I feel very inspired, and will bring a lot of this back also into my work.

00:35:11: would stuck with me is that the bottlenecks how value flows are often not in engineering.

00:35:17: But rather, everything that comes before the design and requirements need to be defined before we actually start implementation.

00:35:24: so shifting left with culture of quality is right thing to do And also having shared ownership over quality and distributing responsibility for this within teams.

00:35:36: I thought it was interesting here.

00:35:39: AI accelerates everything into software development feedback loops, what this does to the processes and people involved.

00:35:48: How we have to unlearn things but that is also normal in a natural process.

00:35:54: I loved the reference of three plus-minus two teams that will be seeing more in software development, having a product owner designer and an engineer.

00:36:04: Plus minus to people working on softer products.

00:36:09: And then last but not least is that we should also not fear because now everything will change.

00:36:14: there's principles that will stay around.

00:36:17: it's very good practice as team to define what stays.

00:36:22: Thank you all so much for tuning in today.

00:36:25: As always, You can find more information about Romano's work In the show

00:36:29: notes

00:36:30: and we'd love For you to comment on this episode where resonated with you or reach out To us through our email podcast at hellofiftyfifty.world If you have any specific feedback Or questions that We should address And if you know someone That should listen to This conversation would Love for you forward it.

00:36:50: And last little question for you to think about is that if you're working in software development, what are some of the roles and skills that your excited when it comes AI not just ones going away?

00:37:05: Thank-you so much for listening.

00:37:07: We'll see each other again soon!