The AI Daily Brief: Artificial Intelligence News and Analysis - AI Adoption Lessons from 5000 Devs

Episode Date: September 25, 2025

Google Cloud’s new Dora research report offers the most in-depth look yet at how developers are using AI—surveying nearly 5,000 professionals worldwide. The findings highlight soaring adoption (now at 90%), major gains in productivity and code quality, and a striking paradox where trust in AI still lags behind usage. Perhaps the biggest insight: AI boosts individual performance, but its real impact depends on how organizations adapt their systems and workflows to capture those gains.Brought to you by:Is your enterprise ready for the future of agentic AI?⁠⁠⁠Visit AGNTCY.org⁠⁠⁠⁠⁠⁠Visit Outshift Internet of Agents⁠⁠⁠Try Notion AI today with Notion 3.0 ⁠⁠⁠https://ntn.so/nlw⁠⁠⁠KPMG – Discover how AI is transforming possibility into reality. Tune into the new KPMG 'You Can with AI' podcast and unlock insights that will inform smarter decisions inside your enterprise. Listen now and start shaping your future with every episode. ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.kpmg.us/AIpodcasts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Blitzy.com - Go to ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://blitzy.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ to build enterprise software in days, not months Robots & Pencils - Cloud-native AI solutions that power results ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://robotsandpencils.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Vanta - Simplify compliance - ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://vanta.com/nlw⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠The Agent Readiness Audit from Superintelligent - Go to ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://besuper.ai/ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠to request your company's agent readiness score.The AI Daily Brief helps you understand the most important news and discussions in AI. Subscribe to the podcast version of The AI Daily Brief wherever you listen: https://pod.link/1680633614Interested in sponsoring the show? [e-mail]

Today on the AI Daily Brief, what 5,000 developers can teach us about AI adoption? And before that, in the headlines, a new video model that's going to take over Twitch. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. All right, friends, quick notes before we dive in. First, thank you to today's sponsors, robots and pencils, Notion, Blitzy and Super Intelligent. To get an ad-free version of the show, go to patreon.com slash AI Daily Brief. And if you are interested in sponsoring the show, shoot us a note at sponsors at AI. Daily Brief.A.I.

It looks like we might have just a very small number of slots opening up for the next couple of of months. 2025 had looked to be completely sold out, but then there was a little bit of jostling and time switching. So if you have a release coming up in October, November, December, one of few shows to spread the word. Again, shoot us a note at sponsors at a.ailydlybreath. Welcome back to the AI Daily Brief Headlines edition, all the daily AI news you need in

around five minutes. There is infinite infrastructure news and just unbridled ambition coming from Sam Altman. And we are going to talk about that, but I wanted to start with something a little bit different just because we have been so infrastructure heavy recently. It's almost like there are a few totally different parts of AI discussion, especially in places like X slash Twitter. On the one hand, you get these big picture, high macro level conversations around trillion dollar infrastructure buildouts and bubbles and all of that stuff. And then meanwhile, as people are debating endlessly, bubble or not, productive or not, you have this legion of creators that are just pouring themselves into

every new model that gets released, discovering capabilities that weren't possible five minutes ago, and sometimes models pop their little head up and suggest that something really different is just on the horizon. This is why I've talked about this idea of a utility score and needing to judge new models on the basis of what new use cases they actually bring online, as opposed to a more simplistic take that just looks at them in terms of traditional benchmarks. Well, one model that everyone is talking about right now is a new video model that's going viral with its incredibly on-point live, deep fake ability. The model can take a video input and a single reference picture and then completely replace the person in the video. After a couple of days of playing, there were lots of breathless

tweets like this one from Siri O Barati who writes, Juan 2.2 animate is crazy and it actually excels at three things from my tests. One, lip syncing, so far the best open source I've seen, beating Runway Act 2, two, consistent lighting and shadows with color tone replication when you swap a character. Three, it keeps the replacement character aligned with realistic body dynamics even beyond the face. It's great for full-body replacement. Justine Moore from A16Z took the model for a test ride and did a bunch of tests and said it's particularly strong at videos where you need to replicate lip sync and body movement. Now she did note that there were some limitations. She said that it needed to have a single character facing forward the entire time and that it works

better when the character is closer in the frame. She also said especially in clips that are over five seconds it can sometimes get a bit off sync with the timing and that in some cases both in terms of the background and in terms of the character, there is a little bit of blending. But overall, people are seeing the possibilities and thinking about what they can do with it. A Google engineer going by Lou writes, seems like a good way for people who don't want to be known to start content creation. We'll check this out. Honestly, scary to do content creation while in tech these days.

Another use case related to that that I've been thinking about. My almost seven-year-old daughter has a bunch of ideas for creating a YouTube channel where she read stories. And I'm much more interested in a version of that where she is an animated avatar character, especially if the source video is still her expressions and motions and all the things that make her unique just without having my daughter's seven-year-old face all over the internet. In any case, if the OpenAI rumors are true, we might be on the verge of another breakthrough video model, and so it's likely that Wan 2.2 might just be a footnote in the larger history,

but we are clearly on the cusp of something very big and transformative. Speaking of big and transformative, the fallout from the Nvidia OpenAI story continues, and boy is everyone involved just absolutely slamming the gas pedal on ambition. Sale Mountman took to his own blog to write a short post called Abundant Intelligence, where he reinforced the themes that he had been sharing in interviews. Growth in the use of AI services, he said, has been astonishing. We expected to be even more astonishing going forward. As AI gets smarter, access to AI will be a fundamental driver of the economy

and maybe eventually something we consider a fundamental human right. Almost everyone will want more AI working on their behalf. To be able to deliver what the world needs, for inference compute to run these models and for training compute to keep making them better and better, we're putting the groundwork in place to be able to significantly expand our ambitions for building out AI infrastructure. If AI stays on the trajectory that we think it will, then amazing things will be possible. Maybe with 10 gigawatts of compute, AI can figure out how to cure cancer. Or with 10 gigawatts of compute, AI can figure out how to provide custom tutoring to every student

on Earth. If we are limited by compute, we'll have to choose which one to prioritize. No one wants to make that choice, so let's go build. Our vision is simple. We want to create a factory that can produce a gigawatt of new AI-e-eat. infrastructure every week. The execution of this will be extremely difficult. It will take us years to get to this milestone, and it will require innovation at every level of the stack, from chips to power to to robotics. But we have been hard at work at this, and we believe it is possible. In our opinion,

it will be the coolest and most important infrastructure project ever. We are particularly excited to build a lot of this in the U.S. Right now, other countries are building things like chip fabs and new energy production much faster than we are, and we want to help turn that tide. Over the next couple of months, we'll be talking about some of our plans and the partners we are working with to make this a reality. Later this year, we'll talk about how we are financing it. Given how increasing compute is the literal key to increasing revenue, we have some interesting new ideas. But if you thought that blog was just focused on the Nvidia deal? Nope, OpenAI has in fact announced five new U.S. data centers as part of Project Stargate. They confirmed that Oracle was partnering to construct

three new sites located in Shackleford County, Texas, Donaana County, New Mexico, and a yet-to-be-confirmed site in the Midwest. Those three sites, in combination with an expansion to the flagship site in Abilene, Texas, are expected to deliver 5.5 gigawatts of data center capacity. OpenAI added that these sites represent a partnership that exceeds 300 billion between the two companies over the next five years. The additional two sites are being developed through a partnership with SoftBank. The first is in Lourdes Town, Ohio and is set to be operational next year. And the second site is in Mylam County, Texas.

OpenAI said that these two sites can scale to 1.5 gigawatts over the next 18 months. For those doing the math, that means they're planning to bring nearly 7 gigawatts of capacity and over $400 billion in investment over the next three years. Now, of course, the big question is financing. And it's not even necessarily clear that Open AI has exactly figured that all out yet, said Altman, I don't think we've figured out yet the final form of what financing for compute looks like. One thing that's very clear is that they are trying to rally ambition, so that financing is just a detail to be solved, not a fundamental barrier.

Writes Brad on Twitter, Stargate will eventually be an American TSM on steroids. This is Sam's plan to rebuild the U.S. industrial base. bringing us back to Earth a little bit, and maybe giving us a little bit of a reveal on how Apple is currently thinking about things. The Cooper Tino Company has added MCP support in their latest operating system update. The developer betas for iOS 26.1, as well as the iPad and Mac versions include the building blocks to integrate MCP support into App Intense. The App Intense Framework is Apple's system for enabling cross-app functionality, for example, allowing various

apps to tap into functionality from Siri, Spotlight, Search, and Widgets. Rites 9 to 5 Mac, that means that based on today's code, Apple plans to let developers used a system-level MCP integration to expose actions and functionalities within their apps to AI platforms and agents. In practice, this means that soon, you could have ChatGPT, Claude, or any other MCP-friendly AI model directly interacting with Mac, iPhone, and iPad apps, autonomously taking actions within those apps without developers having to do the heavy lifting of fully implementing MCP support on their own. Lastly today, from the AI coding sphere, GitHub is rolling out AI coding agents capable of eliminating tech debt. The agents are meant to

automatically modernize legacy Java and dot-net applications, work that would usually take months of tedious and expansive developer time that can now be done quickly and cheaply by AI. Microsoft's corporate VP for the developer division, Amanda Silver said, My goal here is to erase technical debt for the industry. A lot of these organizations have 15, 20, 25 years of technical debt that they've accrued that they can start to take care of in a fraction of the time. Now, this really has been one of the most desired use cases. Back in June, Morgan Stanley showed off their project to update 9 million lines of COBOL code with the help of AI. That project required custom-built tools, and although they claimed it saved

280,000 hours of developer time, it still seemed like a significant endeavor. It sounds like from the announcement that we got this week that GitHub is starting to try to develop generalized tools that can perform this type of task straight out of the box. Now, speaking of developers and how they use AI, that is in fact the subject of our main episode, so here we will close out the headlines and head on over there for all of that. Small, nimble teams beat bloated consulting every time. Robots and Pencils partners with organizations on intelligent cloud-native systems powered by AI. They cover human needs, design AI solutions, and cut-through complexity to deliver meaningful

impact without the layers of bureaucracy. As an AWS-certified partner, Robots and Pencils combines the reach of a large firm with the focus of a trusted partner. With teams across the U.S., Canada, Europe, and Latin America, clients gain local expertise and global scale. As AI evolves, they ensure you keep peace with change. And that means faster results, measurable outcomes and a partnership built to last. The right partner makes progress inevitable. Partner with robots and pencils at robots and pencils.com slash AI Daily Brief. Chatbots are great, but they can only take you so far. I've recently been testing Notion's new AI agents, and they are a very different type of experience. These are agents

that actually complete entire workflows for you in your style, and best of all, they work in a channel that you already know and love because they are purpose-built Notion super users. 's new AI agents completely expands the range of what Notion can do. It can now build documents from your entire company's knowledge base, organize scattered information into organized reports, basically do tasks that used to take days and get them complete in minutes. These agents don't just help with work, they finish it. Getting started with building on Notion is easier than ever. Notion agents are now your very own super user to help you onboard in minutes. Your AI teammates are ready to work. Try Notion AI for free at the link in our show notes.

This episode is brought to you by Blitzy, the enterprise. Enterprise Autonomous Software Development Platform with infinite code context. Blitzy uses thousands of specialized AI agents that think for hours to understand enterprise-scale codebases with millions of lines of code. Enterprise engineering leaders start every development sprint with the Blitzy platform, bringing in their development requirements. The Blitzy platform provides a plan, then generates and pre-compiles code for each task. Blitzy delivers 80% plus of the development work autonomously while providing a guide for

the final 20% of human development work required to complete the sprint. Public companies are achieving a 5X engineering velocity increase when incorporating Blitzy as their pre-I-D-E development tool, pairing it with their coding co-pilot of choice to bring an AI-native STLC into their org. Blitzy is providing a limited time, 30-day free proof of concept for qualifying enterprises. The team will provide a 5x velocity increase on a real development project in your org. Visit blitzy.com and press book demo to learn how Blitzie transforms your STLC from AI-assisted to AI Native. That's B-L-I-T-Z-Y.com. Today's episode is brought to you by Super Intelligent.

Now, one thing that we are having a lot of conversations with folks about is the fact that for some of you, your fiscal year is coming to an end, and that means two things. One, it means planning and thinking about what you're going to do in the next year, and two, it means using up those last of budgets so you don't lose them. If you are an enterprise that happens to find yourself in that situation, super-intelligent would love to help on both fronts. We are moving increasingly towards an annual AI planning model where we map out how you can create an action map of your organization's agent opportunities that represents an executable backlog of

AI and agent use cases that you can deliver on over the course of the next year. Additionally, for those end of your budgets, we have worked out deals with a number of partners where we can pre-lock in general implementation packages even before you figured out exactly what use cases are going to require them. If you'd like to learn more about superintelligence agent readiness audits and this new end of fiscal year plan, visit us at B-super.aI, click get started, and make sure to use the word fiscal somewhere in the description. Welcome back to the AI Daily Brief. Today, we are digging into a monster new research report, and I mean monster. This thing is like 142 pages long from Google Cloud's

Dora Research Program that's trying to answer the question of how developers are actually using AI. One of the lurking questions for some behind all of this AI adoption is, is it actually making us more effective. And while I think that intuitively, the people who use these tools most would argue vociferously and in contrast to any report that said otherwise that, yes, these tools are game changers, people who are thinking about this stuff more systematically obviously don't just want vibes, or at least not vibes alone. And that was why people sat up and took notice in July of this year, when Meter came out with a research paper that suggested that in spite of open source developers thinking that they were more effective with AI, they were actually being

less productive. They were moving more slowly. Basically, in that study, there was a gap between people's perception of their enhanced productivity with AI and what the data was actually showing. Now, I've dug a lot into what I think are the problems with that study. It was quite limited in its methodology. There were only 16 developers that participated. What's more meter's definition of an AI user coming into the study was very, very different from what I think most people would define as a regular AI user. But still, some people found the results interesting and wondered if there was more to the story of AI productivity than the obvious benefits that meet the eye. Now at this Google study, we have a much more comprehensive look at developer patterns,

and because this is the second year that they've done this, we also have a little bit of a longitudinal contrast. And what's interesting is that some of the conclusions and takeaways are very much not restricted exclusively to the software and coding use case of AI, but I think apply for all sort of work-related AI adoption. Now, by way of background, in terms of the methodology here, Dora is the DevOps research and assessment group and has been a part of Google Cloud since 2018. As I mentioned, this is the second year of their AI development focused report. To get the information here, they took in hundreds of hours of qualitative data as well as surveying nearly 5,000 technology professionals globally, and that survey

happened in July of this year. So these results, while not from like yesterday, are still pretty recent. Let's talk about the big banner headlines that they chose to highlight first. First of all, to the shock of no one, AI adoption among software development professionals is now up to 90%. That's up in additional 14% from last year. Now, obviously we are getting to the very top of that, given that there's not that many new people left to adopt, but still meaningful growth between last year and this year. A more significant stat when it comes to this question of, do these tools make people more productive, is that 80% of developers surveyed, that includes, by the way, the 10% who don't use AI, report that AI has increased their productivity. So among the people who are using AI, that number is

even higher. And on top of them just being more productive, 59% also say that AI has positively impacted their code quality. At the same time, there are still big challenges. They write, Our report uncovers a surprising trust paradox. They found that despite everyone using it, there's still 30% of developers that only trust AI a little or not at all, which, by the way, is split between 23% for a little and 7% for not at all. And maybe the biggest takeaway and one that we'll come back to is this. While AI is boosting individual performance, its effect on organizations is more complex. This year's research shows that AI adoption is now linked to higher software delivery throughput, meaning teams are releasing more software and applications. However, the ongoing challenge

remains of ensuring software works as intended before it's delivered to users. And I think if you wanted to sum this up in one way, it's like with any new force, AI is very clearly solving some problems and making people more efficient in certain ways, while also creating its own challenges. The overwhelming sense you get from this report is that the new challenges are a cost that is very much worth it for the benefits that come with this technology, but they are new challenges to be overcome. So let's dig now a little bit more into some of the other things that Dora found. Dora found meaningful increases in individual effectiveness, organizational performance, valuable work, code quality, product performance, software delivery throughput, and T performance.

Interestingly, given how much some people have thought that maybe one of AI's benefits would be to reduce work strain, burnout remains around the same as it did in a non-AI context. What's more in going back to this idea that there are costs associated with all these benefits, one negative thing that also increased was software delivery and stability. When it came to how much devs are using AI and when they started, there was clearly a big inflection point around the release of Claude 3 and 3.5. The median start date for developers in this survey was April 24, with a big spike up in June, July of 2024, which was, of course, when Claude 3.5 came out. In terms of how much time developers are using with AI, the median is two hours, and it is slightly slanted towards the downside,

with the biggest portion of respondents having it somewhere in the one-hour range, and there's definitely a growth in reflexivity and reliance on AI. When asked how often they turn to AI when encountering a problem or task, among the AI users, 39% said sometimes, 26% said almost half the time, 27% said most of the time, and 7% said always. When asked how much they rely on AI, 30% said a little, 37% said a moderate amount, 20% said a lot, and 8% said a great deal. And when it came to the tasks that they were using AI for, 71% were using it for writing new code,

66% were using it to modify existing code, 64% for writing documentation, 62% each for creating test cases and explaining concepts, 61% for analyzing data, 59% for debugging, etc, it goes down from there. But you can see there right at the top, this is not just a tool that's being used to interact with existing code bases, this is absolutely producing net new code. Now, one thing that's revealing, I think, in terms of how far along in their AI journey these survey participants are, when asked how they used AI, only 41% said that they were using IDEs like cursor. The biggest portion, 55% were still using chatbots. Now, I'm not exactly sure what they consider something like Claude Code, but this suggests to me that a lot of this usage is still fairly nascent relative to, for example, the power users that we talk about and quote on this show all the time. And while I gave you the headline numbers on how people perceived it to improve their personal results, the breakdowns are frankly even more impressive. For example, when it came to their

perceived impact on individual productivity, 41% said it slightly increased productivity, 31% said it moderately increased productivity, and 13% said it extremely increased productivity. That's compared to 9% who said it had no impact, and just 3% who said it slightly decreased, 1% who said it moderately decreased, and less than 1% who said it extremely decreased. similar story with perceived impact on code quality. A bigger portion in that group, 30% said that it had no impact. And a slightly bigger group, 7% said that it had slightly worseen their code quality. But 31% said it had slightly improved code quality.

21% said it moderately improved. And 7% said it extremely improved. Now, the story you've heard so far is largely about individual performance. And if you are a regular listener of the AI Daily Brief, you'll know that individual performance is only one part of the larger AI adoption story. Especially in the work context, when it comes to getting these much vaunted productivity gains, organizations have to think beyond just individual worker productivity enhancements and instead think about how they redesign systems to capture those gains and translate them into business signals the market can measure.

And that was definitely a big underlying subtext of the whole Dora report. At the very beginning, Google Cloud says that their key takeaway is that AI is an amplifier. They write that it magnifies the strength of high-performing organizations and the dysfunction of struggling ones. The greatest returns on AI investment, they say, come not from the tools themselves, but from a strategic focus on the underlying organizational system, the quality of the internal platform, the clarity of workflows, and the alignment of teams. Without this foundation, AI creates localized pockets of productivity that are often lost

to downstream chaos. And this is the story that we see over and over in enterprises. Incredibly jagged adoption, incredibly jagged performance, and much of that jaggedness being based on the system. systems and environments into which the AI is coming, rather than the quality of the models or the quality of the users on their own. Trying to go beyond superficial analysis, Dora looked at a set of eight factors to help cluster and understand different team archetypes.

Those factors included team performance, product performance, software delivery throughput, software delivery instability, individual effectiveness, valuable work, friction, and burnout. They ended up clustering these into seven team archetypes, foundational challenges, the legacy bottleneck, constrained by process, pragmatic performers, stable and methodical, high impact, low cadence, harmonious high achiever. These are obviously interpretations of data and reflect patterns that they saw over and over again within the teams they surveyed. Now, part of why this sort of clustering is valuable is to help teams understand what new systems they need to put into place or what existing legacy systems could be holding them up when it comes to successfully

integrating these new tools. Taking, for example, cluster two, the legacy bottleneck, they write, teams in this cluster are in a constant state of reaction where unstable systems dictate their work and undermine their morale. Key metrics for product performance are low while the team delivers regular updates, the value realized is diminished by ongoing quality issues. They find significant and frequent challenges with the stability of the software and its operational environment, leading to a high volume of unplanned reactive work. This also leads to elevated levels of friction and burnout in the team. They found that 11% of the respondents were in this cluster. Obviously, the identification of a roadmap of problems also creates a potential path for solutions.

And in fact, the second part of the report, starting about a third of the way through, so making up a big chunk of it, is all about solutions for these challenges and for better adoption. The TLDR of their whole thrust comes on page 81 where they write, to understand what is needed to scale AI impact from individual productivity gains to organizational level benefits, we need to think about systems. Organizations are less like collections of individuals and tools and more like networks of interdependent parts. Workflows through teams, processes, policies, infrastructure, and shared norms, while individuals.

capabilities play a critical role in shaping outcomes, overall performance emerges from how all these parts interact. To support this, they release something they're calling their Dora AI capabilities model. It's a group of seven AI capabilities that they believe amplify the benefits of AI adoption. Those capabilities include one, a clear and communicated AI stance, two healthy data ecosystems, three, AI accessible internal data, four strong version control practices, five working in small batches, six, a user-centric focus, and seven, quality and internal platforms. Now, pretty soon, I think maybe even for LRS this week, I'm going to be doing a readout of some analysis that we've run on the thousands of executive interviews we've done as a

part of superintelligence AI planning platform. And a lot of the story that we have found is very similar to what's expressed here, both in terms of the challenges and how much it really is organizational challenges that hold AI adoption back, as well as some of these remediation. As I record this, I'm on the road for a keynote. And one of the things that I always hammer is that when organizations are asking what we need to do to adopt AI well, the short answer is everything. It's leadership, its data readiness, its new systems design, and new fundamental thinking. In any case, there is so much more in this report that is beyond the scope of this particular episode. If you really want a very data-rich exploration of how AI is getting adopted inside

developer communities and enterprises more broadly, I highly suggest you check it out. You can find the report on blog.com Google. It just came out yesterday, so it is hot off the presses. Ultimately, I think it's an incredibly positive thing that the shift that we're starting to see in all this analysis is a move away from whether these tools are effective to instead a question of how do we take what are clearly individual gains in effectiveness and productivity and scale them up across the organization to make systems and organizations as a whole that simply work better. That, of course, is going to be the job of the next decade or more. For now, that's going to do it for today's AI Daily Brief. Appreciate you listening or watching as always. And until next time, peace.