Coffee With Digital Trailblazers
From Idea to Impact: Shrinking Innovation Cycles With AI
Hosted by Isaac Sacolick, CEO of StarCIO
This episode of Coffee with Digital Trailblazers focused on exploring why innovation cycles are accelerating and what factors are driving faster development from idea to impact. Isaac discussed three key trends: role convergence where multiple responsibilities collapse into single roles, smaller teams becoming more agile, and AI coding tools enabling faster prototyping and development. The panel of experts, including Roman, Elena, Liz, Derek, Joanne, Martin, and Joe, debated which factors were most impactful, with consensus emerging around role convergence being the most significant driver. The discussion highlighted challenges around governance, quality control, and the need for better risk management as organizations race to implement AI tools. Participants noted that while AI can accelerate development and prototyping, the gap between proof of concept and production remains a key challenge, requiring stronger governance frameworks and more strategic business architecture approaches.
•“AI product owner is part PM, part data strategist, and part technical collaborator.” – Product School
•“Traditional BAs who focus mainly on process flows and requirements documents, without comfort in data tools or experimentation, can feel squeezed between no-code automation, off-the-shelf SaaS platforms, and AI assistants that generate standard documents in seconds. By contrast, BAs who can interrogate data, understand APIs and integrations, and participate in solution design are becoming even more valuable.” – Coursefic
•“87% believe that AI will enable engineers to focus less on scripting and more on system design and directing outcomes. of respondents believe that AI will enable engineers to focus less on scripting and more on system design and directing outcomes.” – Perforce 2026 State of DevOps Report
•“The top 25% of successful engineering teams achieve a cycle time of 1.8 days. 16.2% of teams deploy on demand.” – axify
•“A 5-person team in 2026 can ship what a 50-person team shipped in 2016. AI.
•Smaller teams of 2-5 developers [using AI coding tools] report 68% faster delivery times, while larger teams of 15+ developers see more modest 31% improvements. – Hostinger
•But AI augmented teams need more coordination: 70%+ (i) more horizontal teams, fewer layers, (ii) increased mobility between roles and teams, (iii) blended roles with hybrid responsibilities – Atlassian The State of Teams 26
•92% of US developers use AI coding tools daily. 41% of global code is AI-generated. BraivIQ
•Gartner predicts that 40% of new enterprise production software will be created using vibe coding techniques by 2028. Hostinger
[00:00:01] Speaker A: Greetings everyone.
Happy May 29th.
Happy episode 174 and welcome back to the Coffee with Digital Trailblazers.
Just going to give everybody a few minutes to join. Thank you for keeping this program on your radar and coming back after a week off. And I think it’s important for us to have some weeks off when I know many of you should extend and have a longer weekend. We will do this again.
July 3rd is a Friday and most of our watchers and listeners are in the US so we will have July 3rd off as our next upcoming episode.
But we will be here until then. We will be here through the summer.
Anyone in San Francisco who wants to meet up?
I’ll be in San Francisco next week for this Snowflake conference and then I will be in Vegas at the end of June for the Varant conference. So those are my last two conferences for the season. I think I’ve been a total of eight. It’s been a very, very busy season and part of the reason I went and invested that much time into looking at conferences is just all the innovation that’s coming out there in the form of model improvements and AI agents and new integration capabilities with MCPS and new capabilities with AI agent builders coming from just about every single platform.
Semantic platforms sitting on top of Data fabrics Joe, I’m just going to go off the jargon wheel this morning with all the different things that are happening and what everybody is trying to do is help organizations do what we’re talking about today, go from idea to impact and take the innovation cycles from what was years to then months. And now as I got one quote for today’s opening slide all the way down to weeks.
And that’s just crazy to me.
Today we’re going to explore a couple of different concepts around why our ability to innovate is happening faster, more efficiently, ideally smarter. And you can read some of the information I have here in my opening slide.
There are three things that I see happening that are creating faster innovation cycles and the first is something that was happening pre AI and that is we’ve taken very distinct roles and responsibilities, whether you call it Agile or you call it DevOps or product management.
And those roles are collapsing. And that’s been happening through the appearance of tools. It’s been happening as organizations have gotten better in innovation cycles.
So that’s been happening for a long time. I have a couple quotes in here. Product owners are now part time product project managers, part time data strategists, part time technical collaborators. The BA role is no longer Just about documentation and workflow. They’re getting involved in testing and automation and a lot of different areas.
87% believe that AI will enable engineers to focus less on scripting and more on system design and direct outcomes. And so one of the things that you’re seeing is what used to be two or three distinct roles on an innovation team may be collapsed to one role with lots of different responsibilities.
Now we have what I would just describe as smarter and smaller and faster DevOps teams. This has again been happening pre AI.
The entire DevOps life cycle has been about shrinking cycle times.
Despite all the advances in DevOps, 25% of successful engineering teams achieve a cycle time of just under two days and 17% under 17 teams deploy on demand.
That leaves a lot of room up there. I mean there’s been a lot of talk around DevOps being able to accelerate deployment cycles. But what you’re really seeing here is only advanced tech companies with very cloud native architectures are able to get that kind of cycle. That said, here’s a quote and I why didn’t I have this list? I’ll have to get the reference for you. But a five person team in 2026 can ship what a 50 person team did in 2016.
Smaller teams are able to take advantage of AI coding tools a lot more than our larger teams. That’s because our larger teams still have a lot more coordination they have to deal with.
And again, AI augmented teams need more coordination. This is coming out of a really good report from Atlassian about the state of teams in 26. We’re getting a lot of the tasks done, but we’re not getting rid of the coordination that’s required to make innovation work. 92% of US developers are using AI coding tools probably. There’s a survey that comes out every month updating that number, somewhere between 80 and 95%. But this number was staggering to me. 41% of global code is being AI generated. I have some issues with that number, quite frankly. I just don’t believe it but just included it. But Gartner predicts that by 2028, that’s only two years away. 40% of enterprise software will be congratulated by Vibe coding techniques. That’s just crazy.
Businesses that needed six months and a development team from an agency two years ago now validating ideas in a weekend. And that’s not an exaggeration. This is just crazy to me. Just what’s happen in the last six months in particular around our ability to take ideas and bring them to fruition.
So let’s just jump into our conversation, folks. Do say hello in the comment stream. Do leave me your comments.
Do share some ideas, throw a question out there. I’m hoping to have a real good conversation today. We have two experts that are joining us around innovation.
Whenever we do an innovation topic, I invite Roman to come join us. Roman, you shared some frameworks with us and I threw this first question because it sort of aligned with some of the things you were saying. You know, between role convergence, smaller teams, AI coding tools, and whatever else you want to put into the kitchen soup, it’s reshaping how fast we innovate.
Which of these is doing the most heavy lifting in your experience? You run an innovation lab and which might be a little bit overhyped. Roman, welcome to the floor.
Roman, you are here, but I can’t hear you. Are you speaking? Do you have to go off mute?
And I no longer see him here. He dropped. Okay, sorry about that, folks. Elena, welcome to the floor. Our second guest, special guest, Elena is now becoming a permanent resident on the coffee with digital trailblazers. Elena, role convergence, smaller teams, AI coding tools.
Where do you put your bet on what’s really accelerating and which is overhyped?
[00:07:49] Speaker B: Thank you, Isaac, for the promotion. Hey, I am now official and I actually vote for just about all of it. So I work at a small startup in healthcare tech with a small dev team and we are definitely seeing as we you said about 25% faster or maybe 20, 25% faster shipping from them, which for us in a small startup matters greatly. So I definitely second that. I second the conversion of the roles because commercial roles now have to have enough understanding of AI, prompting AIQC AI workflows to actually even do commercial jobs, let alone doing any technical jobs. But if working in technology, that much more important that commercial teams are now technically enabled. So definitely voting for both of these.
Another one down the commercial kind of line.
The commercialization is accelerating, for instance, prototyping, for instance, positioning and messaging development. All of that is happening with the AI tools. So voting for all of that.
[00:09:12] Speaker A: Wow, that’s handful. That is definitely a handful. Let’s just keep going down the room. I need to ask Joe, can you give me a hand here and send Roman the zoom link.
I forgot to send it to him earlier.
[00:09:29] Speaker C: I’ll see if I have his email. Sure.
[00:09:31] Speaker A: You could do it on LinkedIn. Just do it in a private message.
[00:09:35] Speaker C: Oh, good idea.
[00:09:36] Speaker A: Thanks, Roman. We’ll get you on board in a second.
Liz, jump in here.
You’re going to place Your bet between role conversion, smaller teams and AI coding tools, what’s going to help us accelerate innovation cycles?
[00:09:51] Speaker D: Well, I love the role conversion. I am an entrepreneur at heart and what’s amazing about AI tools is that they are actually enabling this like as you said you can do, you know, validate an idea in a weekend.
The tools themselves are becoming so fantastic that it creates such ease and we can actually move people towards business and understanding their business models, their customers, their strategy, their environment.
So when it comes, it’s almost like back in the day we had like we had these really fifth generation coding tools and we have all this shadow it, but this is actually almost eliminating it and it’s moving us towards true entrepreneurialism, which I just think is fantastic.
[00:10:50] Speaker A: Oh boy. Are we eliminating it Derek? Or you know, we still got guardrails to think about. I’m just, let’s just start with your, the first question. Convergence of roles, smaller teams or AI coding tool. What’s really helping us accelerate today?
[00:11:06] Speaker E: I think the back to your message about working with smaller teams is good, but I think also looking at the coding and the heavy lifting is really coming from the AI coding but also the AI testing with the customers and clients that I work with. I see where they’ve taken now thousands of lines of code and been able to implement more complex coding into that particular code. But at the Same token take 15,000 lines of code which would a long manual process to overcome and kind of work through. They’ve been able to do it within minutes. So now you’re looking at the functionality of amplifying the human capabilities with smaller teams cross crunch functionals. It’s become a game changer. I see where the AI coding tools are becoming more accelerating the process of prototyping the testing and innovative cycles that wherever it needs to be seen and it really help them to, to really look at that and become a force multiplier. When you look at the role convergence from the engineers, the product thinkers, those security professionals and it takes them out of the silo realm. I think a lot of times it and these coders, they’re pretty much just kind of working and doing their own thing kind of in a dark space, but now it kind of helps to find light to what they’re doing. I think also look at just the role convergence is over romanticized because you really can’t take AI and apply to somebody and allow them to be more proficient. They really have to have a core expertise in their domain in which they’re really trying to use it and I think when you look at that, you really can now look at not only looking at creating the speeds to work with it, but also look at the risk. You can’t have faster innovation without the risk. And I think when you look at this from a standpoint of resilience, the speed without enterprise expertise, it really is going to introduce more vulnerabilities faster than we can detect them. And that’s the problem I find a lot of companies have today. They’re creating these new risk and exploits without understanding their current tools don’t have the capability to detect what’s going on. So I think there needs to be a balance. When I look at this, across the board, AI tools will equal acceleration, smaller teams will equal agility, but deeper expertise is going to equal protection and sustainability across the company moving forward. And that’s something that people kind of overlook. It’s not a forward thought or proactive thought, it’s an afterthought. So we need to look at the protection and sustainability of it moving forward earlier rather than later.
[00:13:18] Speaker A: Derek, I agree with the part you’re saying. When you go from idea to poc, I think the coding tools are helping with that acceleration. It’s almost disrupting the old, you know, Miro and you know Balsamiq, you know, ways of prototyping and wireframing. Those are all gone. You could just go rapidly create an app. I also think a lot of the agents that we’re seeing are agents that are doing what automation was doing, but automation plus plus, in terms of being able to do natural language against it, in terms of being able to build your own agents within platforms with a lot of context.
But I still don’t have seen. I have not seen a lot of questions, quote, unquote, orchestrated workflows, which we’ll talk about in another coffee hour. Roman, I think you’re here.
Just say hello. I am sorry. I screwed up.
[00:14:16] Speaker F: Well, it’s a lot easier when you have the magic link, which I didn’t have.
[00:14:20] Speaker A: I’m so sorry.
Just skipping steps. The AI isn’t taking over and saying, you didn’t send Roman the email with the link. I’m sorry, we’re on our first question
[00:14:32] Speaker F: and I’ve been hearing everything, so I’m caught up.
[00:14:36] Speaker A: Where are you in terms of what’s actually accelerating the cycles? Is it one of these things? Is it multiple of these things?
And is there anything that’s overhyped?
[00:14:47] Speaker F: Well, I think Excel, you know, as was spoken before by Derek, AI is accelerating development. And the challenge, in a nutshell is no longer how quickly teams can build something. It’s really now shifted left to deciding what to build, why it matters, and whether or not it creates value or not. Right.
So companies are looking for people that can identify those opportunities and validate solutions.
Figure out how to guide execution through continuous discovery. AI accelerated delivery, which is the coding part we’ve been talking about, and then the end to end product thinking, you know, the guides from vision to validation. And I.
Depending on whether your cycle is kind of a product life cycle oriented or whether it’s an innovation life cycle, which is slightly different, everything’s shifting left.
[00:15:47] Speaker A: And, and how does that make things faster or is it just.
Well, it’s just. Or do you think it’s the same time we’re just doing more up front work on the use case and less work on the construction?
[00:15:59] Speaker F: Well, I think it’s impacting four areas. Speed of execution. Right. And the volume of output. It’s so much easier to generate lots of stuff, including AI slop. The third thing is it’s reducing the cost of iteration. And then the fourth one is we’ve now got non technical people with much better accessibility to technical capabilities. And that’s the vibe coding. Let’s build a prototype in an hour. That kind of work.
Okay. Now the problem is that makes it tempting to ship faster. And how do you know whether you’re really shipping things that are sending you in the wrong direction or not? So what AI isn’t doing yet is it’s not helping you with the quality of the questions you’re asking. It’s not helping whether or not you’re solving the right problem. And it’s not necessarily, you know, telling you whether your metrics, whatever they are, reflect your strategy. It may reflect what your plan is. You know, I got to deliver something by next month. But matching it back to the company’s strategy and having a roadmap to that, that’s still kind of a little bit not there.
[00:17:10] Speaker A: So again, we potentially construct, can construct faster or constructing more.
We have questions about quality and reliability. Martin, this sounds like a CIO disaster.
[00:17:24] Speaker G: Yeah, yeah, that’s exactly where I was going to go. I was going to go with government governance, control.
Whether any of this has value, whether it is actually safe.
And I think the whole thing is a massive potential to really drive productivity, drive business value.
But on the same flip, same kind of thing. On the flip side of that, it’s also a massive opportunity for a big explosion as well.
A couple of other thoughts.
We’re seeing a lot of yeah, I’ve got to be on the AI, you know, kind of hype cycle and everything else.
And that’s driving some really short sighted thinking.
We hear a lot about well, we don’t need some of these junior staff anymore, be it in the business side or in the IT side. We don’t need junior developers for example because we can just vibe code everything.
But the trouble is you need more experienced folk to tell whether what you’re creating and the output of it is actually correct and makes sense and safe and everything else.
And if you’re not bringing more junior people through the ranks and training them up, how are they going to gain experience to look at this stuff in the future? So I, I can see this kind of wall approaching in the sometime in the future where all these experienced people are gradually going to retire or die and there’s a big gap where there was no junior people coming through that are gaining experience.
So you know, are we just going to get rid of the whole workforce and AI is going to run everything?
Yeah. So I can see some short sighted thinking and but the power of this is kind of unbelievable. It’s just how do we get that power and use it right And I’ll give you an, give you an example of the power. So you think yeah a few years back, yeah if you were kind of looking at investing in a company, whatever else you’d pay an advisor or you’d pay a subscription to different companies to get in depth review of a company.
A friend of mine, just for a bit of an experiment asked Claude, gave it a few basic information about publicly traded company and set Claude going and doing a weekly kind of cycle of actually reviewing what’s in the press, reviewing published material, reviewing SEC filings and, and everything else. And Claude built a complete website with graphs of performance and news and everything else in just, yeah a few couple of hours and now maintains this and refreshes, refreshes it from new stuff coming out. That’s the type of thing you’d be paying hundreds or thousands of dollars a month for to a subscription in the past just to give, just use it as a bit of a, an example of what Claude can do just from publicly available information.
[00:20:27] Speaker A: I’m also equally amused Martin, about some of the things that can be built. My, my, my own testing is always been it gets me 80% of the way there but I don’t know how to finish the other 20% and I don’t know if the 80% it built is valid and things that I actually want to use. But you know your point on level one employees, we covered it here a couple of times in the two weeks since our last episode. I was actually at the University of Arizona’s graduation.
Ronan was there. My son graduating with his bachelor’s in aerospace engineering. Eric Schmidt, former CEO of Google, went up there speaking about AI and the boos just kept getting louder and louder.
And so if you go to my blog, drive.star cio.com, couple posts earlier I disrupted my normal schedule to write about this and it is one of the forms of what I would just call AI debt. I have an article coming out around that next week.
Just because you can accelerate doing things doesn’t mean you don’t need to be thinking about the people participating in that conversation. And so I think we’re seeing seeing a lot of this accelerating from idea to poc.
And I’m still waiting for many of the elements of POC to production that we still need to have to see for enterprises to be successful with this.
Joanne, love to hear your opinions on this.
What’s actually driving the acceleration and I know you have some thoughts about going it taking it to the next step from POC into production and impact.
[00:22:15] Speaker H: Well, first of all I would say I do agree with Elena that it is change across the board, the roles collapsing and shift left notions are definitely real. The gaps however between things that are starting to show up and this is around what several other people have mentioned as well. First of all, the qc, the quality and assurance tools are missing. That’s a huge gap. You can’t take for granted what cloud spits out at you as necessarily being true or whether or open AI, whatever, whatever tool you happen to use because the time of learning is different on every tool. In other words, I learned up into this point. Now I’m doing a lot of screen scrape from a lot of different sources and that’s getting fed in etc. You know, I made a kind of pithy comment the other day in a post about do you say you’ve arrived when perplexity quotes you?
You know, I mean like is this the new Google ability you know from from years in the past?
But the quality control issue is a real major one.
You still need seasoned developers and seasoned people to go back and look at the code and make sure it actually does agree. So note to listener, don’t take everything that you read off of an AI for granted.
That’s one area. The other is to the notion of shifting left. What I am seeing a huge resurgence of like we have internally is design thinking, and I mean systems design end to end.
That’s the next step that I think people are going to start addressing because it’s fine to vibe code, come up with an app, test it out, have something, you know, like what Martin was describing, work for you. But it’s that next step after of, well, what’s the next feature and function that we used to do in, you know, road mapping for a, for a product?
That’s the part that’s changing. The business analysts are becoming more directed in what’s the value being created. It’s not just this product iteration, it’s the next product iteration. So I’m seeing a lot of that start to research and what I’m not seeing is a lot of people who are able to do it well.
And that’s a real gap and I think that that’s a new role for people to take on. That’s the homogeneity of the business analysts, the product owner, etc. Etc. Whatever those roles in title actually are, it’s someone who sits down and says we have a strategy, these are the outcomes that we’re trying to achieve. How do I go from there backwards to developing a product that is actually going to fit and not require massive changes every other month or every other, you know, quarter?
Because that’s where the change management issues come in. That’s where so much else starts to really explode for the cio.
So that level of critical thought and systems design, systems thinking, design thinking, those are starting to percolate up to be, hey, we really need people to do this. Some people call this forward deployed engineering, but it’s not actually because engineering is a skill set and this is more about architecture.
So I think we’re going to start to see the emergence of the new AI architect role in a way that brings the business value together with the product innovation and in the event, innovation labs, that’s where I think the next big shift left is going to be.
[00:26:13] Speaker A: Joanne, you and I are in agreement on this one. I think it’s the role convergence that’s the most impactful.
You know, we’ve been, there’s been some of that going on. I mean I don’t have Agile teams with Scrum master project managers, program managers and product managers all sitting together figuring this out. So that’s been happening through tooling for some time. But when you have, when you’re able to do what Roman describes, a lot of shift left thinking and bring that architect understanding, bring that business analysis understanding, bring data capabilities and now Bring top down strategic and customer experience into one one and a half people.
That’s really powerful, you know.
And so now you have like the notion of I think I know what I want to build, why and I’m getting a lot of help on the how that bar has raised pretty significantly and but we haven’t actually seen a lot of applications built this way just yet. You know, I met with, I met with the CEO of Publicis Sapien earlier this week and I asked him that question. I was like, why aren’t we saying more customer facing types of AI applications yet?
And you know, his view of it, which I’ll paraphrase, is that there’s a lot of things happening just below the waterline that will become an explosion in the years to come. And that’s how he described it to me. Go ahead, Joanne.
[00:27:51] Speaker H: Yeah, yeah, so I didn’t mean to interrupt you. I apologize.
The.
I think the thing that is going to drive it is we’re starting to bring enterprise architects into the fold and they’re looking at things from the perspective that will surface that to the customer perspective. What, what I find is really ironic about AI is you have frontier models with massive appeal to consumers, yet the B2B to C or B2C market is devoid of AI apps.
It’s, it’s really interesting to look at it.
We all use these tools on a regular basis, but the get the consumer involved for the mass adoption other than the frontier models is not really happening yet. I think that’s going to start percolating out beginning of next year to the middle of next year. But from that perspective you need to have that enterprise architecture kind of appeal or skill set, I should say, and apply it to the customer side, whether it’s the customer interaction side on a UI or the consumer facing or the consumer appealing application or product that you’re going to put out there.
[00:29:20] Speaker A: Yeah, I agree with that Joanne.
And you know, maybe we need to have a separate conversation around, you know, the AI architect role.
My sense is that they’re, you know, below the line and looking at, you know, which model should they be using and where should they be storing their data and getting into the sort of the infrastructure layers because there’s a lot of questions around that and maybe the encouragement we have to have through this program is that they need to move up into those business tiers and looking at where the, you know, the next competitive value is coming from and work their way back into the architecture. Because what you’re seeing is, you know, everybody, you know is a significant stride to openness.
Right. You can pick any model to work with, any co generator to work with your tool of choice. And so the vendors are trying to lock you into your tool of choice, but they’re creating a lot of openness there. And it’s because so much is changing this quickly. I mean, Carl is asking a question on the common stream about image generators. This image I have here on the slide idea to impact cycle is it’s an AI generated image that if I tried to generate this just two months ago, it would have sucked. I would have never put it up here.
And I just gave it the title and some, you know, maybe one or two more prompts around this and it creates this very interesting image to go. Look at that. Yeah, it’s going to draw attention. And you know, you think about doing that at an application level and saying, how do I design this user experience so people will actually use this? And now the AIs can respond to that. It’s pretty wild. Let me take my break. We’ll bring Liz back. We’re going to talk about governance, customer stakeholders. We’re going to talk about quality and business impact. I have Liz, Heather, Joe raising their hands. Martin is up here again.
Folks, welcome to this week’s coffee with digital Trailblazers. So great. We have great audience here today and just want to give you an update on our upcoming episodes. I’ve actually come up with both June and into July episodes. June 5th, we’re going to talk about growth hacks in the AI era.
Amy Swanson will be back with us. I have another special guest lined up for that one. And you know, the growth hacks was popularized about five, ten years ago as a way to gain traction mostly in B2C companies. I have no idea what we’re going to cover in here, but it’s going to be interesting. That’ll be June 5th. If you’re a marketer, Chief Digital Officer, you don’t want to miss this particular episode. On the 12th, we’re going to take on another side of this idea of going from POC to production. How do we scale?
And that’s, you know, I’m using scale to be very generous, but what decisions that we have to make early upfront to make sure that what we’re pocing is actually going to make it into production, deliver the value it’s promised. We’ll cover that on the 12th.
On the 19th, I disrupted my schedule to talk about this level one issue.
We’re going to bring on a bunch of students who graduated in 2026 and give them some advice on finding jobs and meaning in the AI era.
And I think that’s going to be a really important one for all of us giving back to this generation.
On the 26th, we’ll talk about AI ROI for skeptical boards, what’s actually working in production again, and we’ll talk about how to get our boards on board.
The third, we will take a break for the holiday and the tenth, that was the episode I originally had planned for the 19th, which we talked about reskilling mid career leaders, what senior talent needs to do to stay relevant. So we’ve got a couple areas around career, a couple areas around innovation, POCs and production, and one very important topic around skeptical boards. It’s all coming to you over the next several weeks. Again, remember these two URLs starcio.com Coffee always redirects to the upcoming episode. And then drive.starcio.com Coffee if you missed an episode, I have the episodes there. If you want to see the dashboard, I have them there. And if you want to sign up on Spotify or Apple, I have these going many of the episodes going on to his podcasts later on. So do sign up and join us, however, which way you can. Liz, I’ll bring you back when you can validate an idea in a weekend. What happens to governance like steering committees and sign offs? And how should digital trailblazers involve customer and stakeholder engagement? Sort of two sides of the equation of the upfront work that we have to do before we even start working on an idea.
Liz, which one of these?
[00:34:25] Speaker D: So I’m reflecting back on the comment from your CIO friend from what was a publicist.
[00:34:31] Speaker A: Yeah, yeah.
[00:34:33] Speaker D: And he said how it’s bubbling below the surface or something to that effect. I think that this is exactly the problem is that the governance and the scalability and security associated with actually bringing these products to the forefront, that it’s just not ready, we’re not ready for prime time. And if we, if we talk about, you know, the new role of the architect, it is going to be, it’s almost like when I said before, like the elimination of it, that we’re going to start breaking it down into two primary roles, like the role of making sure it’s whatever you’re producing is hardened and secure and scalable and the role of the true entrepreneur who’s focused on strategy value realization that really becomes the primary driver for everything and the rest of it sort of just, you know, becomes automated.
And that’s what I love like you can really focus on customer impact and understanding, you know, that feedback loop and generating better and better work without focusing so much on individual, you know, we have to make sure everything’s documented. We have to go through the BA process. You have to go through the QA process.
If we can leverage AI to actually improve the security and the scalability, again, that becomes part of the dishwasher. And you can focus on having, you know, clean dishes and having good dinner parties.
[00:36:15] Speaker A: Interesting.
Are we ready for prime time? There’s also a question here in the comments from vavab around cost.
We’ll get to those two, I think when we get to Derek. But Heather, welcome to the floor.
What do we want to talk about here when talk. When we talk about people.
[00:36:35] Speaker I: Well, a couple of comments that I want to make relative to what’s in the live stream, but the first thought that I had was when we talked about how quickly things come developed, sometimes within a weekend. Think about the hackathons that we’ve been doing for a long time. We can create those with the guardrails making like the rules of what is going to have to be done, what has to be considered so that when ideas are coming to be presented, that’s already incorporated. So I’m thinking that we’re not unfamiliar with the short turnaround. There was a comment about the, I think it was, Lucas had mentioned about the assumption whether senior people are learning capable or are spending their time learning AI. I think it goes on two sides of the table. It’s incumbent upon companies to give people the opportunity to learn, to give them upskilling avenues so that they can do it not only work time or through auspices of work, but also in other ways. And then the other side of that, if they’re not going to be training these people and they’re going to just let them go, where does that tribal knowledge go? We’ve talked about that many times on this, on the coffee hour. You’re going to let all this knowledge about your company, the technology, the vision, the mission, the history, is going to just walk out the door. So there has to be an admission by companies to how are you going to maintain this, how are you going to sustain it and how are you going to value it? So I think that that’s important.
One of the other comments that Dana and Joe had mentioned about creativity and what I think people are still going to be there to do is to see the creativity come to life. And how you’re going to interpret what is coming is presented in AI because I Think all too often, many of us, and I put myself in that category. I, I have an idea and if I just put all those ideas in to a prompt, maybe something will come out of it and then I can say yes, yes, that’s what I wanted and I can expand upon that. And that’s the human element that can be ignored. So I think Joe’s Create versus Creative is absolutely an important mindset if you’re going to be using AI in terms of developing things. And then the last comment, and I’m not going to mention his name, but I know he’s on the live stream and he’ll know who he is.
But there are those more senior individuals who are in transition and what are they doing with their time, their free time? You know, they’re looking for a job, it’s a full time job. Absolutely. But they’re also, this person in particular is finding time to learn about AI, taking courses, taking certifications, making sure that you are maintaining that skill set even when you’re not working. Because you know what, you have the opportunity to do it. And that part is incumbent upon you because you don’t have anyone supporting you from the corporate side.
[00:39:41] Speaker A: You know, Heather, I think the, you know what’s happening that you know all the write ups around level one, I think that’s mostly enterprises that are cutting back on that skill set.
It’s partially because they’ve probably overinflated that population in years past. Certainly during the pandemic, there’s evidence of that.
What I think that’s not being talked about is just the opportunity at mid sized companies to be looking for people who have the converged role experience or have the skill to be able to get there. And it’s that latter one that I think is most interesting. It’s like when you can hire that mythical man month in software, that book from 30 years ago, when you can hire that person who can be both a CTO and a chief Product officer and give them that experience on a staff, single user experience to excel at because now they’re working with an AI and a smaller team. I think it’s just going to be very interesting when midsize companies start putting all that together. Let’s go to Joe. Joe, welcome to the floor.
[00:40:47] Speaker C: Isaac, I’ve got a handful of sound bites for you. I’ve been patiently.
[00:40:51] Speaker A: I’m ready, I’m ready.
[00:40:52] Speaker C: You’re ready?
[00:40:53] Speaker A: I’m ready. Go.
[00:40:54] Speaker C: You know, I was out in California and one of the speakers out there threw a new term out and from a Topside view. If you ask what’s driving a lot of the investment in AI, what’s the role of senior management? The board, he called it folly, a fear of limiting innovation.
It’s a fear sort of like fomo, but it’s a little different. I’m going to spend money because we don’t want to get left behind. So to the point about what’s driving it.
Second, I want to talk about we’re going faster, that’s great. But speed without direction is just expensive wandering. Right. We need business objectives. This is, this is everything old is new again. What are we doing? What’s the target? So key roles here are people who understand the business outcomes and can measure in defined ways these how successful the effort is. I read some interesting statistics. I can’t quote the source, but it seems like the expectation is that we’ll be generating lots and lots more code.
And, and we’re hearing that from some of the stories here, but that, that developers were looking for like 24% speed up. That was the. The stat, but the completion time. And it’s funny you alluded to this a moment ago.
Completion time increased by 19%. That was your 80, 20, right? You can get 80% of the way there. The other 20% takes the other 80% of the time.
So speed does not always equal value. I think Martin touched on that and the need for governance. Right. Are we getting the right outcomes?
Do we have faith? Do we trust the outcome? That’s been alluded to.
Liz said something interesting that sparked a thought.
Maybe, just maybe, the role of software engineer in the future will be that of an engineer, not just somebody who writes code, but somebody who actually designs. You know, I thought about like the auto industry, for example. There’s a group of people who design the car and then there are people who turned it into something that can be mass produced.
And there’s a team that watches the quality of the product that’s just rolling off the line. Maybe we need to be thinking about software engineers as being engineers.
And then finally the whole notion of the architect again. I wrote a piece a couple of weeks back.
I think the architect is no longer architecting systems. I think you’re now architecting businesses or you’re architecting the entire organization.
As AI matures, it becomes the tool set for creating businesses, not systems.
So there’s a brain dump across the spectrum.
[00:43:55] Speaker G: That sounded like a mic drop at the end there, Joe.
[00:43:59] Speaker A: I just wrote it down. Role the architect in. Org design Martin, I want you to comment on this last question.
Your initial thoughts, but also what disciplines and changes are needed to make sure speed converts to business impacts and quality instead of just more output. And nobody is allowed to say we need to start out with outcomes or vision statements, which is my version of it. Let’s get into the weeds on this one. Right? Let’s just say we have an idea of why we’re, of what we’re pursuing and why and what the outcome is. What’s next?
[00:44:34] Speaker G: Well, first of all, I wanted to say something else you said about small to medium sized companies.
[00:44:39] Speaker A: Yeah.
[00:44:39] Speaker G: And I have a theory about what is going to happen.
And if you think about small to medium sized companies and what generally happens is they might have a couple of IT people who play with stuff and they bring in some software and they do a little bit of coding and things like this, I can see small to medium sized companies not having IT groups at all, but having business people that are using Vibe coding etc and creating whatever they want themselves.
And this is going to go one of two ways.
They are going to be hugely successful and maybe lucky or there’s going to be some mass blow ups because they haven’t properly checked what they’re creating. They haven’t put governance, they haven’t put cyber security properly around it and things like that. That, So I suspect that we’re going to see a lot of small to medium companies exploding with cyber risk and other problems that they’ve caused themselves. So that’s my, I’m channeling my Derek here. So I, I can see, I can see that happening just on that kind of small to medium sized companies. I think there’s a, A, a tsunami of disasters coming, coming at us, you know, with their kind of liberal use of AI.
So anyway, back to your question.
Sorry, did you want to comment on that, Isaac, before I carry on, I’m
[00:46:05] Speaker A: just going to say I don’t think Vibe coding tools are for citizen developers, business users. I mean I think they’re good at prototyping. If you’re in a walled garden, I would look more at spec driven development which includes a lot more of the artifacts that are classical and software development life cycles. If you look at my blog, the best ones are creating artifacts around data models, around PRDs, around testing, as Joanne mentioned earlier, and they will continue to explode but as they build up more sophistication, we’re going to bring that back into the engineering realm. Like business users don’t want to look at data models for the most part.
[00:46:48] Speaker G: But yeah, you and I both know what is going to Happen in small to medium sized companies.
They’re just going to have people in finance or whatever using whatever AI tool they can and creating all sorts of stuff.
[00:47:05] Speaker A: We shall see.
If you want to create all sorts of stuff and you’re a citizen developer, I do have a recommendation for you. And it’s because the platform that I’m going to mention is it’s called pavement, that platform and others, some others like it are building it on top of existing SaaS infrastructure and so they have natural guardrails built into it. Go ahead, Derek.
[00:47:31] Speaker E: Yeah, I think when you look at this it’s a couple things that come to mind. I mean you just go back to the weekend warrior who’s generating his innovations. I mean they, they think they’re doing something right, but they’re not innovating with governance mindset. And if we look at the evolution of innovation across the years from going from copper to fiber to go from security to non security and add it, anytime you add it after the fact, it creates a problem and it creates more cost. And I think the problem we’re looking at governance today. Governance today cannot keep up with the speed of artificial intelligence. So what happens is what we need to do, we need to get smarter, we need to embed governance upfront. We need to talk about those pre approved guardrails instead of putting them on after the fact. We need to look at automating the policies that we enforce when it comes to artificial intelligence and the development cycles in the pipelines and most of all have realistic risk visibility instead of I think this may happen or static checkpoints. These are things that are not sustainable. And we’ve seen this now with the small companies and mid sized companies and the larger companies that have tried this and have failed. As we continue to move forward, we need to look at, you know, governance should not shut down innovation, but it should enable safe acceleration. And those are kind of things that we need to kind of focus on as we’re moving forward. As I look at the, you know, the speed doesn’t equal impact in the, this discipline thing I look at also we need to start focusing on intentional AI resilience by design, not by reaction. This is looking at security by design, responsible AI checkpoints, accountability, all these things need to come in place. We also need to look at the outcomes of base metrics, not just activity metrics, these are things where now we’re not measuring based on number of prototypes, we’re measuring what’s the business KPIs, how much risk has been adjusted and what’s the value we’re delivering without the risk to the company. Continuous validation. Innovation loops are things we need to look at also where the business plus security equals the customer equals a good product moving forward. Embedding these things up front is going to save us time. It’s going to make it more innovative. But also people are going to be more apt to now move to at the speed they can because now they know they can without the risk of somehow putting the company at risk. The governance piece of it and the risk piece of it. Too many companies too often are overlooking this and we’re seeing the demise of them because of it.
We have to change our mindset, we have to change our perspective. We need to be looking at these things proactively and not reactively. Speed is only an advantage if you can trust it. Trust on what you build. I was looking at security.
When you deploy it, it has to be sustainable and scalable. And these are the things that I think are overlooked. But if we change our mindset, I think we can overcome those. It’s just going to take time to do that.
[00:50:05] Speaker A: Yeah, I mean, I think you’re describing a three horse race between innovation, governance and risk visibility. And you know, looking at this comment from Daniel Giacomoli on, on the common strain says the blast radius is huge and the root cat causes are human, not AI. You know, and that’s really interesting concept because it’s going to be very easy for, you know, for that blast to be blamed by AI, but it’s really, you know, who’s, who’s managing it and who’s accountable for it. We’ve got 10 minutes, Elena, go ahead.
[00:50:41] Speaker B: So I just want to go back to where we started and to your question, Isaac, of the trends we’re seeing, which one is here to see stay? And I actually think the one that’s here to stay is change management because everybody, we are saying elevation of design thinking, the importance of stakeholder alignment, reinforcement of strategy, governance right off the bed and all the way throughout. It’s all about change management and in the open space. Interestingly, change management in AI, it’s not even implementation, right? It’s just exploding. Our business model and where would it settle? Does not get nearly enough play.
So that’s my 2 cents and also a plea to maybe have a separate coffee hour on the importance of change management in the world of AI.
[00:51:33] Speaker A: Oh, we’ve covered that a couple of times and you know, we’ve got a bunch of people here who love talking about it.
You know, I think what you’re really seeing, Elena, is the enterprise response, response to change management in AI is to get rid of the people who are slowing them down.
[00:51:53] Speaker B: Call it AI.
[00:51:55] Speaker A: You know, just unload the ship. We got, you know, 50 people to change instead of 500 and you know, and then those 50 are going to be the, you know, the type A’s, the ones who are early adopters and they’re going to figure it all out. But you know, then I bring back what Derek just said. You know, governance and risk is slowing us down. And you know, we’re talking about this just to bring this all together, right? We have seen a huge accelerant from idea to being able to innovate and create a prototype. We are not seeing a huge accelerant from that prototype into production. And that’s why we’re talking about quality and business impact and change management. Joanne has a solution for us, of
[00:52:38] Speaker H: course I do, it’s called Gale.
You know, it’s, it’s really interesting because one of the things that we looked at very, very early on was governance gaps, execution gaps. And I talk about that being the latencies, you know, my old time to data, time to decision, time to value. And when you look at that they’re really well defined if you look for them. They’re not well defined if you never think about them, them. And that’s where, to Derek’s point, governance needs to be built in from the get go. But governance in AI is really, really difficult because it’s layered. Because you’re talking about dealing with humans as users who have different perspectives, different roles, different security accesses, different intention is something that is also framed by perspective. And so when you talk about governance and you talk about execution, the, you know, my human in the loop versus human on the hook, that’s both ends of those spectrums. And I think the thing that’s going to be around and is really coming out is this notion of, to something Liz said with respect to business architecture and strategy, those forward deployed engineers literally walk around job shadowing everybody else and picking apart processes and seeing where they’re broken and not before they ever touch a keyboard. And I think that’s the trend that we’re going to see come back more than anything else in order to address the governance, the risk, the security, all the things that have to be built in up front. And the notion of the, you know, is it a latency, is it a gap in human to human like an approval process or something like that? All of those things are what, what people will start spending their time looking at more because they can’t deal with the change management issue. If you have a, an LLM that you’re depending on and you’ve wrapped around it and that model changes, think about patching and what that’s going to be like. That’s exhaustive work that no AI can actually help you with. Same thing. If you have a quality issue, go back and trace that code. Yes, you can use an AI tool to help you, but what was the original intent?
Right. In order to get the business value and move from POC into production, all this has to be pre planned.
And if you don’t do it that way, that’s where you fail.
And so the roles are changing. But I think the descriptions of what people are doing in order to design this kind of system needs new language around it because you know, to Martin’s point about change management, something he’s very passionate about, what are you changing?
Are you changing the people, the process, the technology, the vibe coding, the intent, the governance? Are we changing corporate policy in order to accommodate for AI? In many cases we are, are we changing the business model to accommodate for it? So all of this, if it’s not done up front and made measurable return on data, you, you, you will fail.
And I’m seeing it for companies that have done their POCs, moved them into production and, and then they’re still saying now what?
So is it a more expensive fancier dashboard or is it really driving insight, taking insight and turning it into something actionable. And that’s really what the goal should be, make it actionable, make it executable and do it reliably where you have observability, traceability and all those lovely ability words.
[00:56:44] Speaker A: I think we got a lot here going. I got four minutes Liz and I’m going to hear from Roman to close up. Hi Liz.
[00:56:51] Speaker D: I just, I just love this topic and I just love the concept of making sure that we are what Derek said I think is really brilliant, like setting up the upfront implementation of scalability, security, the technical architecture and focus on the organizational change and what the value is for the, the ultimate customer. And I think that that’s, I think, you know, in a few years or maybe even a year because we’re moving so fast, we’ll see this really move in that direction.
[00:57:27] Speaker A: Well, I agree with Derek that we need to have governance and risk visibility, but I’m going to disagree on the upfront part. I think the trick of this is that it needs to be on the side working in parallel with the innovation. There are just too many things that come up. And when we talk about role convergence, one of our issues has always been there hasn’t been a risk management, security, governance person that we can add to every agile team. Listen, CIOs, I think it’s time to do that.
Create your agile teams with risk management as a core function and your ideas may be able to get to production a little bit better. What say you, Roman?
[00:58:09] Speaker F: Oh, I agree. And I guess maybe to bring it all home, I had an opportunity to go to a technical conference by a number of what I would call mid cap CIOs who were successful at AI implementations. Okay. And all of them did what a lot of the people on this call have already talked about about. They started with a problem and a goal and there was one company in particular that was their problem statement was we can’t turn around RFPs fast enough. They have customers coming to them and it takes them weeks to turn around a request for proposal or a quote to these potential customers. They looked at kind of designing a solution with a partner. Partner.
They did it, by the way, for under $130,000. They created a prototype. They learned and adjusted based on that prototype. They implemented it. And in the end, as Elena pointed out, change management was the only thing they got from a 80% success rate. And they kind of got leveled out there. And it was people that were really stopping them from really achieving higher levels of success than the 80%, which I thought was pretty good by the way, that, you know, they went from turning things around in weeks and months to literally a few days.
And the biggest impact was people with tenure, not necessarily people with age.
So that was a big learning for them.
[00:59:45] Speaker A: Maybe that’s the second vote that we have to revisit change management and adoption. Maybe we have to have another episode on governance and risk visibility and a third one around the new roles of architects and organizational designs. Wow, it’s been a mouthful. I feel like we’ve been a little all over the place today. Hopefully this was interesting and useful to you.
Let me know if you have ideas for future topics.
Let me know if you want to meet me in San Francisco next week. I will be there for the Snowflake conference.
And do visit drive.starcio.comcoffee to listen to previous episodes.
Do remember the URL.
Remember the URL for being able to redirect to upcoming episodes. We have one next week around growth hacks in the AI era.
That’s really going to be for our product managers, our digital officers, our AI officers and our marketers. Please do join us next week. We have some special guests for that one lined up. We’re going to switch back to the other side of the extreme from POC to production. We’ll be talking about architecture and infrastructure and scaling a little bit. Touching on people there as well. That will be on the 12th. On the 19th. Advice for college class of 2026 do read my blog post around this drive.starcio.com I have a blog post around how college grads are pissed off about AI.
There’s an opportunity there for every small and medium sized business and I provide some advice for CIOs and talking about that on the 19th.
On the 26th we’re going to make Joe the star of the show AI Roi for Skeptical boards and we’ll bring back the board discussion. We haven’t had one of those in a while. Will be off the third and the 10th reskilling mid career leaders for the AI era.
So that’s our conversations that we’ll be having. Thank you for joining this week everybody. Have a great weekend and we’ll be back here in June for our next episode.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。