Transcript — More software, faster: Odevo's AI Native transformation

⚠ Speaker labels are absent in this transcript. The talk has two presenters: Daniel Jones ("DJ" / "Deejay") of re-cinq, and Tomasz of Odevo (ex-McKinsey). When attributing, prefer "one of the speakers" unless context clearly identifies who is speaking — for example, references to "when I joined Odevo in 2024" or "I worked for McKinsey" are clearly Tomasz; references to "I work for a small boutique consultancy" / "re-cinq" / "DJ" in the first person are Daniel. There is also a brief opening from an unnamed host/MC.

⚠ Speech-to-text artifacts are preserved verbatim. Common ones: "magenta coding" / "a gentle coding" = "agentic coding"; "DOGPT" = an internal GPT product; "loan bull" likely = "Klarna"; "the dopamine in Russia" = "the dopamine rush"; "teacher parity" = "feature parity"; "geo MCP" likely = "Jira MCP"; "ethics" in dev-work context = "epics"; "Ralph Wiggum loops" = unclear coding-agent pattern. Do not silently correct these in quotes — flag them.

---

Section 1 — Opening (MC)

If. You wouldn't. If you wouldn't. Mind. Like scooting. If there's an empty. Chair next to you. Just have some people. At our. Too shy. To go. Into. Content. At all. Okay, amazing. I mean, we have three minutes. Why don't you let these guys get started a little bit early since there's no seats left? So. Impressive. So this is Daniel. And Thomas. These guys are going to be giving you very interesting case study, but they're professionals, as you can see from the not contrasting at all vibes here. Absolute dream team. So take it away, guys. Like AI enhanced face earlier. Yeah. We're good with Photoshop to this. Yeah.

Section 2 — Speaker intros

Thursday good morning here. So, right, I'll kick things off. As you might have guessed, the is about generating more software faster. We'll come back to that theme at the end. And this is really the story of some work that we've done together. One of us is an ex McKinsey consultant, and the other works on a small booty consultancy. I'm not sure if you can tell which one is which. But, yeah. So Dan Jones or DJ to my friends, because the initial was not because of the raving. I'm not actually a DJ or anything like that. And Thomas. Just tell us. What is your poison? Can you tell me you can use one of the five versions of Mini if she wants? To fulfill the publisher? That's an internal sign. Al. Yeah. So I worked for a small consultancy, boutique consultancy. I think he's the most flattering way for that. Come greasy. We're based in northern Europe, and we do AI native transformation. So we come from a background of cloud native transformation, helping people with, you know, cloud kubernetes, all those kind of things. And we're seeing the same kind of patterns around transformation and why adopting a gentle coding practices and getting most out of AI is not just as simple as, like, giving people some poor code licenses. There are changes that need to happen around and alongside that. So we'll go into a piece of advice in this talk.

Section 3 — About Odevo

Yeah. So, yeah, we'll talk about the background. So I'll tell you a bit about over the same page. And we will continue talking about the training, and then together we'll wrap it up with some measurable impact that we have had now. And for people with this. So, yeah, the background, the background of it is a devo is a property residential property management company that has both owns property managers and also provides those services which one of those is technology. Right. And the recent ranking are the third largest tech company, private tech company in Sweden. So if you take loan bull number one, there's another one in the middle. We'll talk about them. And then toss number three. So that's just how relatable it is. And, yeah, we are global, right? We are in the third largest and we have, we offer the full swing. So we own the full entire value chain of residential property management. And property management in this context is like, you know when you're renting a flat and you've got to pay some company like, you know, 100 quid a month or whatever to like make sure all the bins are in good order and the communal lighting and the H track works and the lifts and the elevators and many other things like that. The bills are accounted for and that nothing breaks or if it breaks, it gets fixed. And in the UK, of course, now it's all the fire safety or in the US, it's elevator safety and all the others, everything around it, we own as a deal the full suite of services. End to end in residential property management.

Yeah, it's quite what's up in that like the kind of people Your end customers, your end users. Some of them are volunteers. So they're having good software that works well makes their life a lot easier. They're often under kind of equipped for the job apps and manage finances and maybe the volunteers. So if a devote can do things really well because of large scopes, you make lots of people's lives better and improve the limbic circumstances.

Yeah. Yeah. So we're growing like crazy. We've been acquiring a new company pretty much every week. We have to take into account an M&A in our case is taking a company that's usually a father, a mother, a kid and a dog. There are property manager and we sort of, we buy this at scale across the northern hemisphere from the US to Europe and everywhere else. So we're present in the United States, Mexico, United Kingdom. We originally started in Sweden and then we started growing spending towards the UK, the United States. And the rest of the world. And we're right now focusing in, we're realizing we can grow more time zones. It's becoming a bit problematic. So we're sticking to the ones we have and probably will continue growing either probably north in some parts, south in others, or try and bring everything together. So yeah, it's a massive, massive change in seven years. We've gone from 50,000 homes under management to 2 mil 2.5 million from a thousand employees to 14,000 employees and we continue growing and continue expanding. What as a company.

And those all of those acquisitions and having different solutions in different places means that let's really heterogeneous tech. Kind of situation. So a number of different tech stacks you have like. Now if you have a tech stack, you think you only text it, we probably have it. So we have everything. We even have some of some sort of, we use software that's completely, there's no UI to it and you call a server and it takes sometimes like 30 minutes to just upload information. But it's also very outdated, heavily paper and paper industry. So in the UK, we still have property manager will put how much he has to charge a homeowner association, how much money we post it. And those post intention be lost sometimes. So stuff like that happens still today, which is really cool that the power of digitization in property management is massive. It's everything. There's just a win and there's a case for all sorts of things which will be very important when we talk about the impact. Of this trend. But yeah, yeah. And so there's lots in tech sector because the acquisitions, there's also a lot of different cultures, there's a lot of different processes, a lot of different methodologies. So one way we're talking about a devil, you know, it wasn't so much that there's like one way of doing things and that's what needed to be evolved. There were loads of different ways of doing things.

Section 4 — Starting state (2024)

Which brings us on to like the AI situation where we started this journey. When we, when I entered in 2024, so we had some things in place. We've been already using github co-pilot in our developers were using GitHub Copilot. We built our own chat solution based on open AI, which is a DOGPT for the main reason for that was that we process order private sensitive data, customer data. We did not want it sort of have our employees go wild and, you know, upload information that shouldn't be uploaded to different types of chat. So we build our own solution. And somewhere we were having a bit of clogged island, we would say like there would be one or two devs who got very excited decided to get a license on their own and use cloud or other tools. But it was just sort of like the island here and one guy there, one guy there. Nothing serious. And we weren't really, there wasn't really an adoption of a massive adoption of AI in the devs job. Right. So we only had around 30% and it was the most GitHub co-pilot when we talk about time to merge. Things took a lot of time. When it came to PR sciences was big once a week, sometimes more even less often. And of course the devs were not, it was quite slow.

So we did invest a lot of time as when I joined in 2024 to systemize the way of working around flow around maximizing and sort of the speed and making it transparent because most teams did not even know they were underperforming. They just, everything was kumbaya. Yeah, let's go. We created that transparency which then helped in adopting AI and doing the training to make them more successful. Right. We had to create that knowledge. And that was something that really made the future journey much easier because we already spent a year and a half bringing modern ways of working to the teams across all these tech stacks. All these cultures. So that was an important investment in the beginning. But yeah, we were okay, but it wasn't really like a massive. So a nice squall over.

Section 5 — Disruption coming; the RFP

Yeah, disruption is coming. And we already knew that AI is going to be a big deal. Like this was still pre summer 2025. We again had more and more of these islands, clone islands popping up. And it's nice and flowed out. It's for some maybe. But we saw that it was more and more talk about it. And we realized we had to partner up with someone to help us in this journey to help all the devs. We made a very bold statement that we're going to train every single developer and organization on the same way around AI. So we create an RFP with AI. Those, I mean, all the, we, we sent it out. I made a big mistake of putting on LinkedIn. I'm looking for an AI training company. I got a lot of 15 year olds telling me they have 25 years of experience doing AI training. That was weird. Or of course I had several people who told me there have been no eugenic before a genetic was a thing. So I got around 80 submissions or people reach out to me. I had sort of. Simmered down and tell them to relax. We then we got 10 that we really looked good. We ran several talks with them. And we landed with an idiot with a mustache. That's exactly what we called it. You got to remember his name. What's his name? Yes. Yes. Wearing ridiculous clothes as well.

So that was summer last year and it seems like ancient history. But that was like before five, it was before spectrum development was kind of becoming popular as like September when I think spec kit came out. So you folks were quite ahead of the curve and seeing that this wave was coming. We really wanted, we really, our main takeaway was with AI bad things get worse and good things can get better. So we really wanted to have someone who can help us in that journey. Yeah, it wasn't easy, right? We really did land with three companies in the end that were very close. But again, the mustache. What a week more business get good facial hair. That's one option anyway.

Section 6 — Training: discovery

So let's talk about training. I probably should have given some kind of like photosensitive epilepsy warning. So if anyone is like sensitive to flashing lights or you've got a stinking hangover, definitely not me, you might not want to look at the brightly colored slides. So I'm going to talk to you now about the training, how we did it and some of the things we did before the training to make it more successful. This is not as a sales pitch. This is so that you can go and do this inside your organization and think about the kind of fundamentals that need to be in place to do this well.

So the first thing. Tony would be bright. First thing that we did is a process called discovery. So for folks who aren't consultants, this is where you go into an organization and you figure out what's the current lay of the land. And this is a really important thing to do because of this very authentic door quote might be parameterized. But in the 2025 Dora report, there were a couple. There's one that was, I think, August ish. It made the point after looking at a couple of thousand companies that if you are not doing software development well at the moment and throw a gentle coding at the problem, things are going to get worse. You know, you're optimizing a system, you're making one part of an interconnected system go really fast, you will expose bottlenecks elsewhere. If you're doing software delivery well and you start adopting a genetic coding then things will go much faster and you'll be on a hockey stick of like exponential performance. The tricky part is that nobody knows where this balance point is. Like how crap do you have to be? Is it there? How bad is software development do you have to be to go slower? Like are you in a good place or not? And as an industry we haven't quite figured that out.

But the kind of things that we needed to look at across the various different teams and cultures at things like CICD. Do you have a platform? This is a current bugbear of mine with. I shouldn't tell too many tangents about other cuspids. But if you don't have a platform like you can't ship code reliably and quickly and then you're just going to generate all this code out of nowhere to go with it. Do you have tests? If agents can't run tests to find out they've broken your software, don't be surprised when they break your software. If you're you don't have coding standards. If the humans in your organization don't agree what good looks like then there's a very low chance that an agent is going to be able to produce code that your team is going to approve of. So there are lots of things that are worth looking at and the optimist in me likes to think that maybe this is another opportunity as an industry that we get to look at those fundamentals and go look we really need to improve these.

We really wanted, we made it very clear in the RFP that we do expect to discover recession and we will pay for because we're, I mean it's very easy to make assumptions without understanding. So we really wanted our partner to get to know our company better. We had some bad experiences with companies came in with all these assumptions which were not right. So we really wanted partnering with. This guy to get to know the people that will do the training for to know the company. This was a very important part of making this a success.

So yeah, if you come in on a magenta coding journey in your organization make sure that you're bringing attention to these fundamentals and like raising the flag if they're not good enough.

Section 7 — Workshops (liberating structures & TRIZ)

The next thing that we did was workshops and we got lots of people in the fancy demo obviously which is spotlights old office quite cool be there but doing things in person both to, we did to the issue set training in person but also we did workshops to explore how people felt about all of this.

So hands up if anyone knows what liberating structures are. Paradise, you know what you're just lazy kind of puts hand up. So liberating structures is like a menu of different meeting formats and ways of facilitating conversations that tend to subvert power dynamics and make sure that idiots with loud voices like me don't dominate the entire conversation. So we use a couple of these in person to ask people what do you feel about the upcoming AI training you're about to do. What emotions do you experience? Are you scared? Are you worried that you're satisfaction is going to go? Are you worried about not being motivated in the future?

We also did a round of questions using a format called TRIS which apparently any Russian speakers in the room. It's some acronym for the creative theory problem solving. So we posed a question to the group. Which was actually that you work for a property management company called EVIDO and they're about to hire 200 junior software developers that come from a country and a culture where they always say yes. They never say no they don't push back on anything and they never sleep. They've got an intravenous drip of red bull and they're just going to code all day and all night. How could we make sure that this is a bigger disaster as possible? And everybody contributed their ideas and chatted amongst themselves. Next question was what do we do here that's a bit like some of those bad things and then you see everyone kind of looking at each other going can I say that the testing is not good enough?

Yeah what's really cool it was also unanimous across all these different tech stacks, there's cultures they started saying you do it we also do it. It was like a lot of these like spider-man memes situation. Everyone was like oh my god we don't either. Why? And this is important in terms of reducing reactants, reducing people's hesitancy, reducing the fear that they have, getting them to start thinking about the solutions they're going to need to make this work better.

So this is if you take anything away from all of this really get people on board get speaking to each other face to face empathizing with each other. Find out what their fears are so then you can have a more sensible conversation with them in the future.

Section 8 — Pilot training

And we also did a pilot version of the training as well. So learn all that to get all the little things that could not go right into the big session. So we didn't pilot version that was on November 11 very important date of polish independence day for kids remember 11 11 is post independence day. We did a pilot and that also helped right because we did it in a small group we realized all the things that could go bad or right or whatever to watch out for and then they the big one where we had 80 people much more successful.

Yeah, I learned lots of important things about Poland when working with Tomasz one of which is don't go out drinking with a tall Polish person the night before delivery training or delivering conference talk with you guys to ask me slightly surplus to requirements. Anyway we're already on 15 minutes in so we probably need to speed up.

Section 9 — Training technique: "What did we just do?"

Something about training technique. If you end up going through training delivering an in this yourself or like trying to upskill your peers, a really powerful technique is after just getting somebody to do an exercise is what did we just do? So we would deliver a training module get into an exercise and we'll talk about silvers in a moment but draw up life. Okay what did we just do? Okay so you instill clawed where did that come from? Okay you typed in a prompt where did that prompt go and then it called at all. Okay where do tools actually execute? How does the model decide which tool to use and drawing this up piece by piece means that you can leverage different types of mental machinery we have. So it's got social importance because you will remember sitting there awkward and going oh I've got to answer the question because you do must session to ask it. We're not going to go on until we've answered people contribute things funny things happen. It's a visual way of representing stuff. So it really cements people's knowledge in a way that's just doing the exercises doesn't so I would really recommend trying to use a technique like that when you're maintained in person.

That was also a massive thing. We really brought everyone into the office in person. We fed them of course they gave them fe count which is a big thing. We fed them, we then had an after party everything around it. But by bringing them into the office telling them the first half of the day no laptops. We were going to talk it out. It also made the training stick more right rather than them being of course doing multitasking and not listening to the guy with the mustache putting them in the office to see each other, to meet each other to put a oh this is all you, it's you okay that also made a message that we pushed for that heavily that it has to be in person it has to be soft.

Section 10 — Syllabus

So the syllabus then in terms of what we actually taught people really fundamental thing teaching people about context management and maximum effective context windows. If you don't understand that the more you add into your context window more off the rail to the airline is going to get then people start doing daft things like adding every possible instruction for every possible scenario into their agent's MD which has been shown in a couple of academic studies now to have negative impact. You're better off having no instructions in your agent's MD then lots of them for example.

We also did exercises with old models that are easy to get to hallucinate and to do propped biasing to demonstrate people the failure cases. Go into standards. We don't need to train people on how to use coding agents on a happy car. It's when it goes wrong. Like when they work it's magic and you just get working code out at the end. So the important part is teaching people about failure modes and that's what the first module was all about.

There's a big jump between using something like vscode github copilot and called code for a lot of people psychologically. Like when you're in vs code you choose which files are open and you can look at them when time you like when you're using claud code in a terminal then all of a sudden the agent is doing stuff and it chooses what to show you. For some people that was quite a big jump. So that's a thing to be aware of.

We talk to people about MCP and spectrum developments and eventually go on to multi agent workflows and things like gas town, now gas city. Which by doing all of those things they then had the building blocks by the end to think about how this would be applied to a team process not just to an individual.

Another thing that we did that was quite important is we did not deliver this training with one gun. So there's a journey that people need to go on and we did weekly half day modules and then at the end it's like okay you just learned that MTP why don't you go away back to your desk, install the geo MCP ask it like can you figure out what ticket I should do next? What do you think are the acceptance criteria and then allowing that to be pushed back and find out whether this stuff works in real life not just in a training exercise and make sure you tell IT ops that people will be installing software on your laptops we lost like 15 to 30 minutes. We're not training because people couldn't install docker. That was great.

And another important part of this psychologically is people like the training exercises gave people a guided tour you know like here's core code here's how skills work. But it's when they then go back to their own desk to make their own minds up and get the chance to play with it on their actual production code basis. That's when people start to see the power of this kind of stuff rather than just when we're showing it to them.

Section 11 — Impact metrics

So should we get to the impact? So of course there's impact. Massive everything up everything all the good things are all the bad things down but yeah what it really did we do see after we finish these trainings in February so we train the whole entire group within that then I did a little more with others but in general by an admit that we were finished with our core group of developers.

And we are seeing massive. We are seeing great things right so in general we're seeing AI adoption now around 94% based on did you use AI in the last 10 PRS 94% said at least in at least three or more 60% in 10 out of 10 and then a bit less for only 6% of organization did not use it and it's more interesting why rather than do it right so now we're doing interviews with that group on this what's happening why aren't you using it to make your life and the general like what's what's blocking you?

We are seeing massive increase in time to market so we are able to push things faster. We are able to build things smaller and better quality and of course yeah it's it's also the whole merge and releasing things makes it much easier. So we're seeing that and if you look at the pure throughput then it's increasing right so it's it's transparent we also build a series of dashboards and every desk can see every other devs where they are working which repos they're working on. We're making it super transparent and calls the org across all teams. We reinvested in that as well so we have this data on a click of a button you can even chat to where you can have conversations. It was very easy to find what's working, what's not working and of course we're seeing constant increase in it.

I think you know giving people tools we would probably see similar trends but I don't think they would be as impactful. And that was the training that enabled that they're more intentional how they use it and they use it I think in a much smarter way. So people are not you know pumping everything onto open is 4.7 just because they aren't experimenting they're playing with things.

Section 12 — Boldness as the real metric

But you know all these metrics are great but I think the biggest impact of our AI transformation was in this. This is this you know you can if you find enough smart people in a room hopefully all of them will start maybe not as exponential in this case but there should be an increase now the tools are getting better what we saw if you go to the next slide was the boldness. People are not afraid to try things. I think that's the biggest takeaway for me if anyone asks how do you measure a successful AI transformation? It is how many more bold things are you doing? How many experiments are you trying out? Are you actually trying to do a full end-to-end agenda workflow?

So one of our goal we did a very bold move we took a platform that we've been building for eight years wasn't a great success what we did is we took it and we rewrote it. Into it rewrite it took us with the support of an external expert but we build a future parity one to one in three weeks. So something that we built for eight years we were able to rewrite in three weeks teacher parity one to one and now we're continuing to build it have a team dedicated much more than we had before so that was 20,000 building it for eight years now we have four devs building it with with AI that that was a bold move trying to build a mobile app in three days we try that. Did it work out? Not so great but we tried they were doing a lot of also if you remember when I showed you the map right all of that like that fragmented software all countries are doing different things is also because of regulation of legal stuff everything is very complicated to build a single piece of software for all those markets but now we're trying we're going to try to build a dedicated platform that will be able to tackle all the all the needs of all the markets in a similar source in a single place and then of course customized with AI for their different needs but that could have never happened last year or without this feeling of boldness.

Section 13 — SDLC reinvention (one team's AI-first redesign)

We got four minutes and 45 seconds left so with this in particular so we did the agency coding training people were using cord code on an individual basis. One of the teams that so much mentioned took a step back and was like how do we reimagine our software delivery process like working on individual stories just doesn't work when you can go so quickly with coding agents so what they did was an entirely different approach that was AI first they spent two weeks talking about product requirements speaking with their males having everything automatically transcribed from which they then went on to get agents create a product wireless document and they really matter tested that you know they thought it through very deeply they sketched out entity relationships on the business logic get that domain really well defined then ship it off to agents who split into like a task plan kind of like spectrum development but like upper level of abstraction if you like and now developers pick up a couple of ethics a day and if one developer wants to use B man sure fine if the mine wants you spec king fine some people are still vibe coding but they each have their own way of working once they kick up an ebb and they're no longer doing human code review so they will look at the code themselves they're not writing code anymore on that team but they will do like seven or eight passes of a genetic code review before then raising a PR which they will then self merge so they don't have the bottom neck of other humans reading code like humans doing stat analysis of code is a silly idea on birthplace so they've gone. Fully AI netted in all of the steps and trying to use the existing scrum Kanban methodologies is you're going to hit bottlenecks very very quickly so once you've got those practices in place then it's time to reassess your SELC and how you go about doing things.

And they're sharing this right so they're again sharing we're making it more like we're using them as a lighthouse everyone can go and join talk to them see how they're doing so it's also a massive transformation we've also talked about this we do a podcast workshop has been on it and so one of the developers on this team so the latest episode is all about this process methodology.

Section 14 — People stories (Dominic, demo team dev, Nicholas)

Speaking which the impact on people so this is the Dominic who would try to pronounce that last name how has he done authentically environmentally very nice publishing he doesn't speak gold it has a hard time yeah yeah so he was an AI sceptic when we came to the training he had previously got chat TV team to write code in isolation you know non agentically it was like this is rubbish it's never going to work and also like quite legitimate ethical concerns about AI wanted trains the energy usage and all that kind of stuff it was when he had the opportunity to play with Opus 45 back in November and freedom to explore that that's when he realized. That the previous beliefs weren't true and that when running agentically these tools can create good codes that he would be proud of.

You will find in a lot of organizations people that trade vs code get on co-pilot like a year ago and it was rubbish and they're like oh I think it's never going to work. I joined it once it was crap and it's never ever going to be good you will find people like that and they need the opportunity to see the current state of the art in order to change their minds and like don on the podcast was saying I haven't written any code at work for four or five months what was interesting is he's doing it for fun at home, you know just like you might buy furniture from Ikea but then do carpentry in your spare time that's what the craft has turned into which is part of our demo team we have a dedicated team that focuses on developer experience this also hasn't written code and he's building all the tools to improve the developer experience in our organization so they build like massive amount of tracking and telemetry and informational tools and in doing that all without wearing he was one of the more one of our like top skill top performers I think he said he's increased his performance by 2000 oh shit

the final calling that I want to mention is Nicholas we invested in a train the trainer program from the start we made it very clear we want to have build this capability to train on AI and then develop our training continuously evolving. Internally so nucleus became a trainee who was also a high performer already heavily doing things with he was also been educated on like Ralph Wiggum loops and all that he's always been a heavy user of AI by becoming a trainer trainer now goes to our to the remaining sort of companies that we still have developers that weren't part of the initial one and training them telling about his experience and that makes it much more real. I mean mustache and tracks look and all yes but if it's a real person that understands the industry can share their own real experience for working at a Devvo that makes it even realer right that makes it more people some people don't like to be top editors but we didn't need to target

Section 15 — What's next (product, then everyone)

so the future and the way that we the future we have now shifted. We shifted we realized that by investing in our engineers we've forgotten about another very important part of the organization which is our product colleagues designers VAs product managers so now we're taking the training that works very well for engineers and we shifted towards we've adjusted it we pivoted a bit for a product colleagues and you'll find that the product folks become the next bottleneck we've seen that in lots of different places. Has I think you mentioned in the podcast so this is and then the next step is once we turn our product managers we want to now train an AI and how to use basic things with cloth code and others to our remaining employees which is pretty interesting it's 13,000 people that we need to educate on basic use of AI and we're seeing people already using it. We have financial managers or accountants using Claude code to build local small apps for themselves so we want to help them structure that with trading as well got three more to line so we will get lunch soon

Section 16 — Cautions: compulsion & overwork

and then yeah and what one thing you've also noticed I hear more of that today I heard like we are having some of our engineers saying I can't turn it off. It's 6 p.m. 7 p.m sometimes 1 a.m the dopamine in Russia is real so now we're also having sort of discussions with them like hey put it away go enjoy life go outside I mean a lot of sun go outside I would love to tell you about skinner boxes and psychology behind it but it's a real the compulsion is a real thing also this has happened in the devos case but Lauren Pete CEO of multitudes did some research that showed that in a lot of organizations people are working overtime because of the increased feature pressure of you've adopted a genetic coding you should be getting loads more stuff you should be getting lots more done and people are now actually doing overtime because that perceived pressure.

Section 17 — "Everyone a builder" vision

Our goal and our ambition is to make everyone a builder we want to we want to pivot from software we built for for our opcodes to opco's build software with our support we're making that shift we really believe everyone in the Devvo will be a builder and we really want 14,000 people to be building stuff for themselves and in the past like in the 70s if you wanted a new business to process right you've got some paper you made a form you told people about it that's how you change the business then our software developers got involved and leverage the power of computing like we had to devolve all of that to engineers and then you couldn't make any changes yourself and it's got worse in the cloud era when everything was distributed complex now we can re-empower business users to solve their own problems like some of the product managers are using Claude code now revive coding and logging board and other tools as well and building things that are used every day so yeah what we see is not that software development is going to go away not that we're going to end up with software developers needing to be made redundant because of those bigger boulder changes that people are now empowered to make we see a world of more software that is more bespoke more usable solving individual people's business problems rather than kind of big slow software those complicated Excel models that used to be someone in the house's hard drive were moving that to buy covid solutions that are usable used by actual users in organization which is super exciting. But also a bit scary.

Section 18 — Close

So yeah a fun journey ahead of us if you do we give questions or we run over that time for questions I'm afraid yeah sorry lunch at this very soon checking the head if you can support me I might be hardly mindful afterwards there's our LinkedIn QR code but other than that I'll shut up and say thank you and you'll be great more detail. Ed there's plenty of food don't worry that'll be fresh.