Now Shipping: A Mind the Product podcast

Inside the world of AI Agents with Dan Olsen

• Mind the Product

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 36:14

Dan Olsen is a product management consultant, educator, and author best known for his work on product-market fit. A veteran of the Mind the Product community, he runs hands-on AI workshops for product teams across industries — battle-testing tools so busy PMs don't have to. In this episode of Now Shipping, Dan joins host Mike Belsito to dissect the biggest AI story of the past month — Google I/O — and examine what the rapid expansion of agentic tools means for product managers navigating an increasingly automated world of work.

We discuss:

  1. How Gemini Spark, Google Stitch, and Antigravity collectively signal that Google has moved from playing catch-up to serious competition with Anthropic and OpenAI
  2. The clear arc of the agentic arms race: OpenClaw showed what was possible, Claude Cowork made it accessible, and Gemini Spark is Google's bid to own the personal AI agent space
  3. Why running agents in the cloud — not just on a laptop — solves real reliability problems for knowledge workers
  4. How the bottleneck in software development is shifting from engineering to product management, and why that makes strong PM judgment more valuable, not less
  5. Why vibe coding and agentic tools increase the temptation to skip discovery and rush straight into solution space
  6. Why product sense and product taste are the new differentiators — when anyone can build anything quickly, what you choose to build is what matters most
  7. Why investing in hands-on AI learning is no longer a nice-to-have for product managers

Referenced:

SPEAKER_02

I'm Mike Belcito, and this is Now Shipping, the weekly show where we dive into the AI news stories that matter to you as a product person. And every week we'll be coming out with new episodes highlighting the important stories for that particular week. But once a month, we host guest experts for interviews to dive deeper on one of the big news stories that month. This month's feature is with Dan Olson, author of the Lean Product Playbook. I had a great discussion with Dan. Let's get right into it. Dan Olson, it's been a while since we caught up. It's always great catching up with you. Thanks so much for taking the time to join me. My pleasure, Mike. It's great to be here. I look forward to talking to you today. You know, it's funny. I used to have an interview series online. At first it was called Product Lunch, and then they were called Industry Interviews. And I believe you were my very first guest for that interview series. And now with now shipping that I'm bringing on guest experts once a month, I figure it's pretty fitting to have you on as my very first guest for the show, too.

SPEAKER_00

I appreciate it. I'm excited to be there. Yeah, I remember that.

SPEAKER_02

I still have my shirt, my t-shirt from the with this show in particular, what we're trying to do, this whole concept is really honestly for my own personal needs. It's been hard for me to catch up, Dan, like just on everything that's going on with what's coming out with AI and not just what's coming out and what's happened and you know what new releases have happened, what you know, what announcements are being made, but what's relevant to me as a product person. I mean, and that has just been genuinely hard. So I created this just so I could get caught up, and then I figured let's share it to the world. I could do it in real time, and everybody else could benefit as a result. So that's kind of the purpose of this. But for our conversation, I was hoping we could just take one big story and just maybe dissect it a little bit, maybe go a little bit deeper.

SPEAKER_00

Definitely. And I think just to yes, Ann, what you're saying, I think everyone's feeling overwhelmed by the pace of change of AI. You know, I teach a lot of workshops and give a lot of talks, and people like, I'm so busy with my day job and my family and my life. Like, how do I have time to like first off stay on top of all the AI news? Secondly, let alone get my hands dirty playing with any of this stuff. So I think it's I think you know, having discussions like this where we can kind of summarize things and like separate the noise from the signal, you know, and what do people really need to pay attention to? I think it's super valuable for product, busy product people, like all the busy product people watching your show.

SPEAKER_02

Yeah, 100%. And maybe before we jump right in, I do want I know that you've been, you've been really digging in yourself. You've been experimenting, you've been like obviously for the longest time, you've got given workshops and you've been teaching and advising. But can you just quickly share with everybody how you've been digging in lately in this new AI everything world?

SPEAKER_00

Yeah, I mean, I basically, you know, obviously used ChatGPT when it came out like everybody else and played with Image Gen. But then about a year ago, when vibe coding started to become more capable, that's when I really kind of jumped in with both feet and saw what was capable, what was probably possible. And the big thing there is actually even I calling it AI prototyping, AI prototyping really for product teams that wanted to create prototypes and test their ideas before they built them, they knew that was the right thing to do, but they just didn't have a design resource or a prototyper around. It really just unlocks them and unblocks them. So right. So that's the cool, that's what I saw. Because when I'm talking to groups, I ask people usually, hey, who who raise your hand if you feel like you have enough design resources to prototype your product ideas? And usually it's like less than 10%, often less than 5% of people raise their hands. So for me, I saw this huge value of hey, you don't have to be technical, you don't have to code, you don't have to be a designer. You can go from your PRD to like a concept that you can iterate and show people and get feedback on. So that's why I've been working on that. Those tools have come a long way in the last year. And then most recently, now we're going to talk about today the agentic world, like with like things like cloud cowork. Now it's unlocking like task automation, which is a you know a whole different use case, a wider use case, frankly, than just AI prototyping. So, like if you're a knowledge worker, there's all kinds of things you can be doing with cloud cowork to first off be more efficient. And I basically delegate the grunt work, you know, like I call like type one work is your busy work, type two is your thinking work. You know, spend less time on type one so you can spend more time on type two. So I'm excited to get into that today.

SPEAKER_02

Yeah, I love that. I love that. Well, let's let's get into it. When I was looking at the you know, just the stories that have happened over the past month, and you and I were trading emails on, you know, hey, what's actually happened? What can we dig into? We saw one big thing that was coming up while we were emailing, and that was Google I.O. Of course, we didn't know what exactly would happen at Google IO, but Google I.O. has has come and gone. Now we've we've seen that, we've seen the announcements that were made. Um I'm curious what stood out to you. I because they made quite a few announcements, and I know everything from you know what's happening with Gemini Spark, which you know, it's right, Google's foray into a Gentech work, um, and lots more. But yeah, what in particular stood out to you?

SPEAKER_00

Well, it's funny because I just kind of put it in the arc, the overall arc of you've got this like AI arms race going on between like OpenAI and Anthropic and Google. And I think that you know the narrative, it's funny, it's not even history, it's like recent history was like, hey, Google was behind, right? And they you know originally had BARD and they were doing demos and they were embarrassing demos, and you know, but basically when Gemini 3 Pro came out, they finally kind of surpassed uh Chat GPT that caused a co-red at Chat GPT. So the bill, I think the big story is just hey, Google's in the game, and with this I.O. event that they just had, all the announcements they did that show that they're they're definitely in the game for real. So, as you mentioned, the Google Spark agentic things that they did. Also on all the other fronts, like they Google Stitch is uh the vibe design world in the last couple of months has gotten really interesting too. So Anthropic launched cloud design, which takes advantage of Opus 4.7's increased visual capabilities. So, you know, you got cloud design, you've got Canva with AI. So everyone's kind of getting into the into the design world. And so Google Stitch uh it's one of the top tools that I use with people. You know, I just battle test these all in my workshop, so I have no no reason to favor any tool except for which one seems to work best. So if you're not a designer, it's pretty cool. So they advanced stuff there on the vibe coding front. They also did a new anti-gravity, which is their IDE. Again, they're late to the game. Everyone was using cursor or Visual Studio Code. Now anti-gravity is there, Codex is there with their app, Claude has their apps now. Everyone's kind of it's funny. I have my vibe coding spectrum, and in the early days, people started their bets in one lane. Like we are either a CLI tool or we're a desktop app or we're a plug-in for an IDE, and now everyone's got an IDE. You know, they want that desktop surface area. So it's just really interesting to see the competition that that happens there, and Google announced a bunch of things, which is which is pretty impressive.

SPEAKER_02

Yeah, well, let's talk about some of those things, and because you're right, it's it's not just one thing, it's the whole totality of everything. And it's really sort of Google, I feel like, coming to say, like, hey, we're very much in this too, you know, right. Maybe, maybe it took us a little bit to get caught up, but but we are here now. For for Google, for Gemini Spark in particular, you know, this is I guess I I feel like the conversations move from again before at first we're all talking about LLMs, and now it's it's this agentic age that we're in. It's actually like agents that can do the work for us, you're right, right? We tell the agents what to do and and they can go off and do the work. And and that's where Google was lacking a little bit, but now with Gemini Spark, especially getting MCP support, so now you know they're you it can live inside of the tools that we're using and and vice versa. Like it can live inside of the tools we're building as well. When you saw the announcement about Gemini Spark in particular, uh the 24-7, you know, personal AI agent running in the cloud. Like, what stood out to you most?

SPEAKER_00

Well, again, I just go back to how did we get here? And I think that narrative arc starts with OpenClaw when it came out, just showed the world what was possible with a gentic. And it's funny because we got this in the AI world, the memes and the buzzwords changed so quickly. Like at first, it was all about prompt engineering. It's like, no, no, that's dead, it's context engineering. Now it's like, no, no, that's dead, it's agentic now, right? So agentic is the the word de jour, right? But basically it started with open claw showing you what can an agent do, but you know, to be honest, like you had to kind of be pretty geeky and technical and go get your own computer and like go tinker and get it. It's like it's like a you know, it's like a Jedi building their lightsaber or something like that. Like the number of people that were so into it and had the technical skill to do it was pretty low. Um, a lot of people did play around with it. And so I think that the next uh part of that chapter is anthropics saying, like for me, if you look at the release pattern that they did for Claude Cowork, right? It came out in December as like an MVP, but clearly in February, March, and April, they were putting out releases, releases, releases. My external view was they had a checklist of all the open claw features and they were ticking them off and trying to get feature parity, right? It's not exactly the same, obviously, but you now have this agentic workflow you can do inside of Claude's ecosystem, which a lot of people trust. It's an enterprise secure system, right? I'm still running on your computer with your files, obviously, and there's different controls you can do. But for me, Cloud Cowork was anthropic's response to open claw, and now Gemini Spark is basically Google's response to kind of the agentic open claw and cloud co-work game. And so I think that's the thing is everyone's now placing their bets on this agentic workflow. One of the interesting things is, you know, I've been running these workshops, and it's a lot of people get kind of confused between cloud chat versus cloud cowork versus cloud code. It's kind of confusing, especially if you don't see the cowork tab. Because the only way to access cowork is you have to be on the desktop app and you have to be on a paid plan, right? So it's like some people don't even know which one they're using or they switch between them all, right? So it can be very confusing, right? So but basically, cowork, the thing that sets it apart is it works on your computer so it can access your file. Now, some of the people in my workshops are always like, Dan, like I migrated to the cloud. So one of the most interesting things for me, Mike, in the announcement was that they're letting you for Gemini Spark use dedicated cloud virtual machines, right? And so I think that that's there's this like interesting, it's like, okay, are we gonna people mainly gonna do this on the desktop? Are they mainly gonna do it in the cloud? Are they gonna do it in a way that's synced across it all, you know, so so that it's like more mobile and you know, you can kind of just take your work with you wherever you go.

SPEAKER_02

Yeah, yeah. And I even, I mean, I I know one of the announcements is like, hey, even if you are using Gemini Spark on your laptop, by the way, you could it we it could be locked. Your Mac can be locked and things could still be working. Yes, it feels like it's been giving it's giving people a lot more freedom in how they can actually run the agents that they want to run.

SPEAKER_00

That's a great point because in my workshops, people really start realizing we talk about scheduled tasks and they're like, oh, so does my laptop have to be open? And uh yes, your laptop needs to be on, powered on, it needs to be can't be asleep, it's got to be connected to the internet, and you have to have your cloud app open. And so, yeah, so like it's funny. I ran into this the other day. I have a few scheduled tasks that do things for me and cowork that I really enjoy. But I was traveling and I got to my hotel room late and I didn't bother connecting my laptop to Wi-Fi before I went to bed. And so, sure enough, that night the scheduled task didn't run that I needed because it wasn't connected to the internet, so yeah, it totally kind of from a 24-7 availability, being able to have this run in the cloud really kind of you know, avoids those issues for sure. Yeah. The other interesting thing is they called it a personal AI agent, right? They made it a point to the 24-7 personal AI. So there's also this interesting like use case thing. You know, if we take a step back again, Chat GPT, in a sense, was a broad B2C application, right? It was just like, hey, it's just kind of like Google search. Anybody can ask anything, do whatever you want. And then, you know, Cloud, and then you've got Cloud Code, which is more like clearly for developers. With Cloud Cowork, we're trying to, it's like, it's really kind of more of a work thing. I mean, you could use it, obviously, you can use it. Like I, one of the first things I did is I when I go to a new town, I want like I just went to Pittsburgh, I missed you there. This skill that would basically, whatever town I go to, it'll like research the restaurants based on certain criteria and create a spreadsheet, right? So that's an that's a more of a personal example, I guess. But a lot of the cloud enterprise stuff, people are using it for work stuff, but they're these Gemini Sparks being pitched as a personal AI agent. So it's very interesting to see who's going after the work use cases and market, who's going after the personal use cases and market.

SPEAKER_02

Yeah, well, and it's almost like we because you mentioned before, like when uh when open claw came out, you know, that was definitely you had to be technically you had to have some technical chops. I mean, hey, I I I have my own open claw like myself. I it I don't have a lot of technical chops, being honest with you, but I it's a it took me a little bit to like sit along and go back and forth and like, all right, help me figure this thing out. And finally I did. Do you feel like it's almost like yeah, with Claude Cowork, it's appeal it's still kind of you know, business use cases people are using for a lot, but to your point, you could use it for personal use cases. So what Gemma with Google's offering with offering with Gemini Spark, do you feel like it's trying to appeal more to just sort of like I don't want to say the everyday people quite yet, because but yeah, but I think it's feels like we're starting to get to that point.

SPEAKER_00

I think you know, if you take a step back, it kind of makes sense because what is Genai and chat displacing? It's displacing Google search, like search, and who's the king of search? Google. So it kind of makes sense they want to protect their broad consumer offering. And so pitching this as you know what I mean, as an extension. In fact, one of the things that they also did is they relaunched search, they redid search and it had funny little side effects. I always love the UX side effects when they do stuff. So, like it people were noticing, like if you Google with the new Google search, if you Google disregard, it interprets that as an agentic command, like forget what I just told you to do, instead of actually like looking up the word disregard. So you went into little things like that, right? So I do think it's deliberate on Google's part that they don't want to lose that broad, that broad B2C applicability that they have. So it kind of makes sense for them to start there to protect their flank, if you will, right?

SPEAKER_02

So yeah, 100%, 100%. Well, I guess one other company we haven't talked yet about, but that is also like playing in the space is Microsoft. Maybe you could talk a little bit about that because they've made some announcements as well. They're starting to collaborate in particular.

SPEAKER_00

I know it's super interesting because you know, obviously they were an early investor in Open AI. In the early dates, it was like, cool, copilot just uses open AI. And in the vibe coding days, you know, because Microsoft owned GitHub, GitHub Copilot was one of the most popular kind of plugins for Visual Studio Code. It kind of made sense that they had all the components together: Visual Studio Code, GitHub, and OpenAI to do the copilot, right? Um, the interesting thing I was really interested because I actually learned about I don't, you know, a lot of the companies I work with tend to be more Google Suites, but I was uh when I was in Pittsburgh, a lot of people there in financial services and they were Microsoft shops. And so I was talking to one product leader and they were telling me how, like, yeah, now I can use co-work in the browser. I'm like, oh wow, that's like interesting because it was specifically designed to work on a laptop. But so it's called, it's funny, it's a mouthful, Microsoft Copilot Cowork. And in my head, as they're telling me this, I'm like, in order for this to have happened, Microsoft had to go cut a deal with Claude, and they had to figure out how to make this work with like you know OneDrive instead of a hard drive. But sure enough, they did all that work and they did it, right? So um so now you can use Microsoft Copilot Cowork as part of 365, you know. So yeah, they I think that they again they don't want to miss out. You they're they're they're hats in the ring as well with OpenAI, Anthropic, and Gemini, and sorry, and and Google, right? So they don't want to miss out either. So it's interesting to see how things play out.

SPEAKER_02

Yeah, for sure. Well, I'm again, I'll I'll think about the last like 30 days or so. So obviously, like again, big announcements at Google IO. I I think all of us, you know, this will be more and more relevant to all of us as we start to play with all the tools. That's the thing now. Like now it's not I at least this is my take on it. I'd be interested to get your take. I I'll just ask you right now. It's like it used to be that you kind of learn and master one tool and you get good at that tool, and that's what you're gonna hang your hat on. Now I feel like for product people, you have to learn about all the tools. You have to at least get your feet wet with all the tools to at least at least just have the knowledge and experience of it. I maybe not be an expert at everything. I don't know if you can be an expert at absolutely everything, but things change so fast and the tools are changing so fast. Do you feel like it's important to at least get your feet wet with these tools, or what is your take on it?

SPEAKER_00

I think you have to get your feet wet. I think, and what I the funniest thing is I run so many vibe coding workshops, and the biggest value add of doing a workshop is frankly is creating space and time for the learning and the hands-on. Yeah. Because you're so busy, like you'd have to be a very proactive self-learner to be like, you know what, it's Friday night. I'm gonna go play around with some vibe coding tools, or I'm gonna go play around with OpenClaw or some agentic AI, right? You know, a lot of people just don't have that spare time. So when when I do my corporate workshops, the first and most important thing is hey, we're carving out five, six hours for just to learn this. And then the second thing is just kind of onboarding them to the tools and then and like just getting over the initial hump of okay, this is what you do. You take your PRD, you put it in here, you paste it, and you hit go. Second part is when you get the first prototype up, now what it's like, well, is there something you don't like? Tell it how to fix it, right? So I I just I just repeat the Nike slogan, just do it. Like if you're like wondering, hey, I wonder if this just do it, just try it because nothing can hurt, right? So to your point though, I don't think you need within a category, like I wouldn't say, should you have tried AI prototyping? Yes. Do I need to try three, four, three or four different AI prototyping tools? No, you can go with the generally prevailing wisdom, I think. And there's wisdom out there of what people you know have used. And that's where again my vibe coding spectrum has different categories, but like in the non-technical PM category, like a lot of people use lovable, and it's like you know, pretty much the go-to for new people in that category, right? If you want like uh if you really want an IDE, then like cursor has historically kind of been the one in there, you know. So and same thing with the task automation. As of right now, I think cloud cowork has the lead, right? So if you're just gonna use one, that might that's that's the one that people are doing and using right now. But I do think it's important. To your point, though, it's not it's definitely a lot user-friendlier than open claw, configuring open claw. But I still feel like you know, when you're doing cloud cowork, you're trying to build a skill, you're trying to build something. You do often get into kind of some debugging things that a little technical helps. We're not quite there where anybody who doesn't know anything can get exactly what they want. And there was a great tweet, actually. Aaron Levy, the CEO of uh Box, who I consulted to back in the day, had a great tweet the other day. He's like, all these CEOs, he's like, they have this AI psychosis because like they play around with it and they're like, oh my gosh, look at this thing. But there's this, like, what do you call it? And I believe it, I see it, the last mile problem. Yeah, you might get that thing working 80%, but to make sure it really works reliably and handles those other cases and things, there's time you got to put in. There's like a QA phase. I hate to say it, but there's a QA phase. And to get through that QA phase, sometimes you have to be technical, right? Like the other day, I was building, I was building a skill in Cloud Code because I got sick and tired of my inbox being like, you know, report spam does not create the proper feedback loop. So I was like, you know what? I'm gonna, for my new cloud cowork, I'm gonna build an inbox thing where I can have a blacklist and a whitelist, and I can just block people so that you know, anyway, it was great, but it's sitting there in cloud coworking. It's like, okay, here's your Python, here's your Python script. Copy and paste this into the terminal, right? And that's fine. It kind of serves it up on a silver. You don't have to know anything, you just copy paste, you do it. But at the end, after doing a few times, I'm like, can we not just make this a web app? And they're like, Oh, yeah, sure, we can make it a web app, you know, with buttons. Like, why am I copying and pasting like command line stuff, right? So things like that, just knowing what's possible. But I think they're gonna keep getting better and easier for for more and more people to access.

SPEAKER_02

Yeah, yeah. No, that it all makes sense to me. And uh, it's like first first things first is is get started. But yeah, the deeper you go with something, then the more you could learn about it. And yeah, being technical does help, but it shouldn't prevent anybody from at least no, I wouldn't I definitely mean you're in learning, right?

SPEAKER_00

Not at all. And you again, there's plenty you can do without being technical. It's just like if you get it's kind of like being in the deep end of the pool, you're like, uh oh, all of a sudden you run into some choppy water, it can help you get out of a pinch. It's just it's just silly that as powerful these things are, it still like makes bad decisions. It's like, oh, I know, I'll do it this way. And then you go to you kind of go down a dead end. It leads you, it sounds like it knows what it's talking about, like it sends you down the primrose path, like, oh, I know this is how we'll do it. And you get to the end, you realize it's the dead end. Yeah, and like we should have took a we should have taken a different turn like 15 steps ago, you know. So it's kind of funny when that happens.

SPEAKER_02

That's right. All right, well, I'll go, I'll go back again third past 30 days. So we talked a little bit about again, Google I.O. But there's so much more that happened. So yeah, at this point, now Microsoft Agent 365, it's generally available, that's available to everybody now, right? GPT 5.5, that this has become now the default. There's a gentic capabilities built in, like it's it's taken a huge step forward when it comes to helping people on on the agentic side of things, not just you know, chat GPT. MCP, MCP now, I think it's over 97 million installs. Just like we it if we're talking about total installs, and then of course now like Gemini Spark coming out of I.O. Many of those big stories. It's the the one common thread is they're all in this agentic world that we're talking about right now. It's like agents, this is the conversation that everybody's having right now is how can we get better with agents? Even I mean, I guess the stories nobody likes to talk about the stories where we hear companies cutting employees, you know, going through layoffs. Some of that, whether it's right or wrong, it's we point at oh well, now we have teams that can manage agents better than people that like Right. We we want to become an agentic workforce. Right. Which includes humans and agents. Right.

SPEAKER_00

The other interesting thing is yes, there's cloud cowork, which you can use in the desktop app, but Anthropic is also rolling out these general purpose agents as part of their platform. And what I'm seeing them do is start to actually obviously you can create your own agent, which is helpful and interesting. But why should you have to create an agent if it's like a well-defined workflow? And so what they the first vertical they started launching these vertical agents, it was in the financial services space, like a financial analyst. Oh, if you want to do some financial analysis, just use our financial analysis agent, right? A little bit after that, they launched a bunch of small business agents, which were like, hey, you want to do accounting? We can do accounting for you. You want to do invoicing, we can do that, right? So I think that we're gonna see them continue to do this uh like vertical by vertical in whatever order makes sense to them, like rolling out agentic things that you can just grab and use as opposed to having to configure.

SPEAKER_02

Yeah. Yeah, well, then that's sort of the next step I wanted to talk about is like all of these things that are happening, like and including all these things that you just laid out too. What does it actually mean for product people? I mean, yeah, like I'm thinking about it's funny because a year ago, I and all of this technically existed even a year ago, but like in in early in early forms and not with not with all companies putting it out there, but right I I remember the talk was like some companies like, hey, how do we prevent bots from being on our website? We're seeing a lot of bot traffic, bot activity going on. We're trying to optimize against the bots. Now we need to optimize so that our website is bot friendly, right? Like we need we need to that's one big thing I feel like that all of this, like again, if if the theme of last month and honestly it's just gonna be every month, is agentic work. It feels like that might be one big takeaway for product people is how do we make sure our products are actually agent-friendly. That's one thing that comes up for me. I what's your take on that though?

SPEAKER_00

Well, yeah, no, I think definitely, you know, people are talking about like instead of SEO for websites, the agentic equivalent of that, right? Or AI equivalent of that, which you're already seeing agencies and people, as you see commerce. The funniest thing is anytime you have a platform, people are gonna want to advertise and sell stuff on there, right? And so that's the thing is like you're seeing commerce kind of come in. Walmart's done some interesting things there with that. But uh, so I think that's one thing that certainly can happen. I think the main thing is like you were saying, MCP usage is it's funny because in a sense, MCP is like a fancy name for API, you know. And now the cool thing is like in Cloud, they've kind of dumbed it down even more. They just call it connectors. It's like, hey, you have a Gmail connector, you have a Google Drive connector, so it's not some like scary MCP API term. But the bottom line is you're gonna want to string together data and tasks from different applications as part of your workflow, and so you need to be able to programmatically access those so that you can do the workflows that you want. Interesting tidbit there, for example, is the official Gmail connector that Cloud offers when you go in there. I was like, cool, I'm gonna do for my inbox scrubber, I'm gonna use that. Well, turns out we this is the dead end, and we start going down the path, and it's like, ooh, I just realized that this this connector doesn't let me create filters in Gmail. And we need to create filters for what we're envisioning. Just because it comes from Cloud or it's official doesn't mean it's fully supported. So then it's like, okay, well, we can use the Google the Gmail API and to get this done. I'm like, cool. So we just switched, we pivoted to that. But that's an example of like a dead end that you would get into. So yeah, so more and more, there's going to be more interconnectivity to let you do stuff. Um, that I think that's that's really important. And uh and you know, the other thing too is like Claude has a skill, has like an MCP connector creator skill. It's funny because a lot of people they don't even realize there's an official one, and so they just create one. They just have Claude create one for them, even though there's an official one sometimes. So yeah.

SPEAKER_02

Yeah. No, I that it all makes sense. And yeah, it's I I will say one thing on my mind is that it it feels like now, especially with the advancements that we're seeing with the Gentic AI, it's easier and easier and will continue to get easier and easier to build. We're building faster than ever. It's not all these, you know, these companies that are making their announcements saying, hey, we're cutting because we don't ha need as many people because we're gosh darn so efficient now, right? Like all the agents can help us be just as efficient. It feels like for product people, that that's a great thing, but also I wonder what that means for like, are we building the right thing? And you, I mean, gosh, you wrote the book on finding product market fit. One of the books, one of one of my favorite books on that. Yeah. Um, I guess what would be the 2026 version of it? Like in the world that we live in today, like what do you think about you know the fact that yes, we could build faster and faster. How do we make sure we're actually building the right things faster and faster?

SPEAKER_00

Yeah, no, it's it's an interesting thing. And I think the way one way to look at it is just, you know, we've always talked about the trio or triad of skills that every product team needs, you know, like a PM, UX design, and dev, not to say other roles like QA and other things aren't important too, but you know, those are the three that people focus on. And we've always had like these ratios that are, you know, and that's the beauty of being a consultant. I can see all these different companies, I can see these guys have this ratio, this company has this ratio. And so you these rules of their general rules of thumb, but they make sense. Like if you have way too many engineers per PM, there's no way the PM can keep up. And there's no way they can do discovery, there's no way they can keep the eye on the customer. They just have to, they become like a jirojockey just running all the sprint ceremonies, and they don't have time to do that. Right. So there's ranges that let PMs kind of do their job. And I'm not the one that commented on this, but I'm seeing this too. But like Andrew Ng specifically is commenting on like as the engineering portion of gets more efficient, you can think of it like a factory line. It kind of reminds me of the book The Goal, where you got a factory line. Like, what is the bottleneck in the factory? Historically, the bottleneck has always been engineering. Our goal is to make sure that backlog is filled so that the engineers always have something to do. Well, as they become more productive with these tools, Andrew Ng and other people are seeing the bottleneck shift up to product management, which is okay, if it's quicker to build, now we're getting rate limited on what do we actually build. And so, I mean, the good news is PMs are required for that, right? And back to your other point, when it becomes, as I like to say, when it becomes easy to build anything, then what's gonna determine whether one product succeeds and one fails? And of course, it's all the foundations of product market fit. Like, do you really understand your customer? Do you understand what problems your product is supposed to solve? Are you solving the most important problems? Like, do you have a unique value prop, how you're solving those problems in a way that's better or different, right? And so I think that kind of human nature, the temptation is if it's easier to do something, it's gonna get done more, right? And you know, I just I just saw someone who like had a posted a photo on LinkedIn and it had the little had the little Gemini watermark in the bottom right. And I was like, hey, just so you know, that's telling everybody that that was made. Like, and they're like, oh yeah, like so. It just if people, you know, if it's easier, they'll do it. And so I the big part of what would what I talk about is problem space versus solution space. Gosh, vibe coding makes it super easy to just go rushing into solution space even more. And so I think that the the most people feel like that good PM skills, and it's funny because we have new terms now that we didn't have a few years ago, Mike, right? People are talking about product sense and product taste, right? And these are not things that anybody used to talk about at all. And I think that is cat that those terms are capturing what we're talking about, which is what's that judgment of what's going to really create value for customers versus not that you need to bring to the table when it's so easy to build anything, right? Also, the building is is fast. There's another important part, which is after you build it, getting feedback and iterating based on that feedback. That is a separate skill that you can either be slow at or fast at. And so I think again, you know, strong PMs that are fast at that can really have a much faster flywheel of building, learning, you know, and iterating than other companies.

SPEAKER_02

Yeah, and it's not to say that people should slow down completely, but it's don't skip over those loops. Like you're just mentioning, it's kind of a loop, right? It's like it is a thing, and then you're getting feedback on the thing, you're learning, and then you're iterating. Yeah, we can't just speed run it, so we're skipping over those steps. We could we could go through that loop a lot faster now than we could have a decade ago, which is awesome. Right. We don't want to forget that loop, too, right?

SPEAKER_00

Right. But also, I'm just a realist. If it's if it's like, oh, because we're understaffed in PM in this new world, oh, hold on, just to time out, I need to do my discovery. No one's gonna go for that. It's like, no, sorry, the train's leaving the station, whether you're whether you're on board or not, like we're shipping stuff. So part of it is uh realizing what the bottleneck is, trying to advocate for you know more PMs or a ratio, whatever that ratio is, you know, whatever the healthy ratio is that's there. So the other interesting thing is like what's going on with design too. The design is kind of getting a little squeezed out in some companies, I feel like, because it's like, you know, because we can create stuff, we can create designs. Now, I would be the first to say I value UX design a lot. When I was a first brand new PM, I realized I knew nothing about UX design, how important it was. Those tools are not gonna necessarily create anything like a 10x designer is gonna create. But to your point, I mean, it's the point of like people are gonna take the path of least resistance. If it's quickest to just be like, hey, let's do this, then it can be hard. Again, the you know, the design people might be in a situation where it's like, wait, hold on, I need time to think about the design. And the org is like moving so fast, or like, well, we'll get back, we could get back to us later. We're gonna ship this and then figure it out. So it's it's a challenge, like just to take a stick back. Like as the pace of work accelerates, as the development productivity increases, it's just a challenge for organizations to adapt their ways of working and their staffing levels. It's just that's just a challenge.

SPEAKER_02

Maybe we could close with this. If there's one big takeaway that you hope people can get from everything that's happened over the last just say the 30 days. Again, some of these stories we've been talking about. What's that big takeaway you want them to walk away with?

SPEAKER_00

I think it's what you said, which is hey, you know, there's there's a whole there's several agenc solutions out there now to help you automate your work. And uh and it's worth it to figure out how to carve out time to at least play around with these tools. Because, you know, again, like we haven't talked about it. The economy's not that great. I don't think PM as a career or function is threatened per se. I think if anything, and there, you know, in the early days of AI, people like, oh, AI is gonna kill PM. And I wrote a blog post saying why I didn't think that was the case. I think we're seeing it play out that no, if you're a really good PM, you're actually in more demand because they people need that product sense to inform what they're doing. But you need to basically invest in your learning and you know, and and and and play around with these tools. So I think it's it's you know, I think I guess a one way to think about it is it's never been more important for PMs to like take charge of their own learning and human capital development in the age of AI in the past. Like, yeah, it'd be great if I took a design class, or it'd be great if I learned how to do Figma. It was kind of like more of a nice to have, but I feel like now it's really imperative that if you want to stay up to date and stay, you know, kind of uh keep your skill sharp for today's marketplace, that you need to take the time to do it. I know it can be hard. Um, one of my resources is actually YouTube. So I have multiple email newsletters that I that I get, but also on YouTube. Um, there are people, the second there's a new feature or something that comes out, they do a quick demo of it. And so it's a quick way to grok. If you're already up to speed, it's a quick way to grok the new stuff.

SPEAKER_02

Yeah, I was just gonna say, like the beauty, because I think you're totally right about now product people, you have to take that beginner's mindset and just start digging in. And the beauty is we have the whole world at our disposal online. We have access to more ways we could train ourselves than ever before. And it's right there, and most of it's like it's free. We don't even have to pay for these these things. So that's the that's the good thing. If for people that are willing to put in that work and willing to learn, the world is our oyster, really. It seems like, right?

SPEAKER_00

Yeah, and if you want to learn in person at mind the product in Chicago, I'll be teaching an all-day workshop on this stuff. I'm really excited for that event.

SPEAKER_02

Yeah, I'm excited to have you. That's right. We have our Mind the Product Chicago event taking place on October 7th, but on October 6th, you could come for a full-day workshop with Dan. And I I mean, I've put on many conferences in the past and have hosted you, Dan, for many workshops in the past. And your workshops are always the highest-rated workshops we'd usually have in that year. So I have no doubt that people are going to get a lot of value from your workshop. Maybe do you want to give just the 30 second um what it's all about? There could be whole new tools that come out between now and then, right?

SPEAKER_00

So right now, the But I know you're gonna update it by then. No, I and I'll update it, right? So, you know, I it'll have a little bit of kind of AI prototyping in it, a little bit of vibe designing in it, and then uh, as of right now, a fair bit of cloud cowork task automation in it, right? So it's a bit of kind of all the top things, the tools that AI-wise that you need to learn. And again, if you're having trouble carving out time and you want to invest the time and have someone you know, experience teach you about it. And I think that's the other thing, is just my learning and teaching, it comes from the foundations of PM for my book. So it's not like I'm just some developers coming in here showing you how to code this kind of stuff. It's grounded in PM and real PM use cases. And like, I don't expect you to code or be technical or to be a designer, but how can you use these tools? So, as of right now, that's what it is. It's vibe, it's like AI prototyping, some vibe designing, some cloud cowork, but I'm sure between now and then, well, one of those tools might get bumped, or we'll add one or two more tools. So it'll be dynamic, but it'll be it'll be long story short, it'll be an in-depth, hands-on, you know, you'll get your hands dirty doing real work with the latest AI tools that are relevant to product teams.

SPEAKER_02

Yeah, that's great. Well, I'm excited for it, and I'm excited that you were able to join me today. So, Dan, just thank you so much for being here and just sharing everything that you're hearing in this world and you're learning on the fly. Thank you for sharing it with us. Thank you, Mike.

SPEAKER_00

It's always awesome to talk with you, and I hope this is useful to product people out there.