Hope you enjoy this episode!

As an aged father of little ones, I can’t write AND podcast at the same time. My sleep-deprived brain just can’t handle it.

So yes, the written version below is, in fact, AI generated based on what I discuss on the pod. Thus, the best experience I think is the audio, but in case you can’t experience that, a written version is below.

Let me know if you enjoy it by subscribing below, I hope you do! See you next week…

Subscribe now

Slides are also here for visual reference.

Hey everyone, and welcome back to the Robotic Scholar Podcast. Week 51. We’re so close to Christmas and New Year’s. I’m excited; things are slowing down, everyone is getting ready for the holidays, and I honestly really like this time of year. It’s a beautiful time where meetings get canceled and people are taking stock. I hope y’all are enjoying some downtime as well. It’s been a year for me, that’s for sure, and I’m appreciating the break.

The 3D Printer Ejection

One thing that has been making me upset, though, is that I am still in 3D Printer Hell. I thought I got out of it—I even posted on Substack Notes that I finally got the printer working—but the errors just keep coming. I’ve decided to just eject on this printer.

I put the situation through Gemini and asked about the Creality K2 Plus relative to the printer I probably should have bought from the beginning: the Bambu Lab A1 Combo. People seem to love Bambu Lab; they’re like the Apple of the printer space, while Creality is like Android. I have nothing against Android—I love the OS and how it works natively with Google—but I do not want to be training to be a 3D printer mechanic.

I got into this to print robots and shells for my Jetson Orin Nano; I’m not looking to get deep into the mechanical process and swapping out parts. Every time I fix something, something else breaks. Now it’s an MCU error. I don’t even know what that is. I found a Bambu Lab on Amazon—it’ll be here Saturday—and I’m paying someone on TaskRabbit to get rid of this hundred-pound Creality machine Friday afternoon.

The Creality did do some great work, though. I have this R2-D2 leg shell it printed for the robot I’m building. So, shout out to the K2 Plus for that, but we are done.

Gemini 3 Fast & The Platform Advantage

Let’s get into AI news. I am thrilled that they shipped Gemini 3 Fast this week. Their timing was impeccable because ChatGPT 5.2 had just released. I was starting to miss that “polish.” The voice mode on Gemini was leaving a lot to be desired, and being in “thinking mode” all the time was getting old because queries took two or three times longer.

I got Gemini 3 Fast on my phone Wednesday morning. It’s great. I’m not a ChatGPT hater, but the power of these solutions is in the AI plus the platform combination. Gemini clearly got the memo internally to use Google Tasks for storing tasks. It works natively with the products you already use.

I have a hard time seeing how ChatGPT competes with that without building their own email, tasks, and notes apps. You need deep integration to provide the experience Gemini offers out of the box. OpenAI is trying to keep it lean, which is smart in the long run, but how long can you do that while Google is nipping at your heels?

Riding the Frontier Wave

Regarding the product I work on, Assist at Procore: we just did a foundational model upgrade to the latest frontier model. There’s a great article called The Bitter Lesson that talks about how it is so much better to ride the wave of frontier models than to think you can engineer your way into better responses.

Because the new model has been iterating in the background and becoming more agentic, it was night and day at selecting tools and understanding context. We didn’t have to do much; we just upgraded the model. If you’re working on an AI product, it is critical to stay on the frontier. You are better off riding that wave of capability than thinking you can engineer your own capability better. You can’t.

AI Won’t Save a Bad Strategy

The last point is that AI with a poor product or business strategy can’t save you. You can bolt AI onto a product, but if that company is set up to fail in this new world, it won’t work.

The online learning industry is a great example. Before AI, you learned from books and 60-hour video courses. Then AI happened, and the utility of that went away. Why watch hours of video when you can just ask a chat interface your question about Python and get the answer? That business strategy was destined to be bulldozed.

It’s like when Excel shipped. The “bean counter” jobs went away, but analyst jobs rose up. If you aren’t willing to completely recast your product strategy to be additive with AI, you are doomed to fail—slowly, but ultimately. No technology in history has saved bad product decisions. Look at Google: they realized they hadn’t executed on chat, so they changed their whole strategy, reinvested, and now they are arguably one of the best providers of AI models and platforms. You can’t just bolt it on and hope. You have to put the thought in.

On that note, I hope you have a great rest of your week. I’ll do one more episode before Christmas. Enjoy the soft week.

Hope you enjoy this episode!

As an aged father of little ones, I can’t write AND podcast at the same time. My sleep-deprived brain just can’t handle it.

So yes, the written version below is, in fact, AI generated based on what I discuss on the pod. Thus, the best experience I think is the audio, but in case you can’t experience that, a written version is below.

Let me know if you enjoy it by subscribing below, I hope you do! See you next week…

Subscribe now

Slides are also here for visual reference.

Siri’s AI Revolution & My 3D Printing Hell

We’re back, baby.

It’s a couple of weeks before the holidays, and I am so excited to give you an update on what has been rattling around inside my head. We have a lot to cover—from Apple’s potential AI comeback to a disaster in my workshop—so let’s hop right into it.

🍎 The Case for Siri (Yes, Really)

I’ve been thinking a lot about Siri. Now, before you balk and say, “Oh my God, Siri? What a waste. What a flop. Apple is so far behind,” hear me out.

I agree. From a user point of view—and honestly, from a fan point of view—Siri has been a disappointment. I was so excited when Apple bought Siri back in the day. It had so much promise as something forward-looking. The idea of a voice assistant driving the entire Apple ecosystem was so clutch, and the fact that it just didn’t happen is a letdown.

What we are seeing now with ChatGPT Voice Mode is exactly what we all imagined Siri would be.

I’ve been a Product Manager inside a number of tech companies, and I know how these things go. If you aren’t viewed as the strategic priority, you don’t get the funding or the love. And if you look at Apple’s meteoric rise over the last 15 years, it hasn’t really involved Siri. It’s been about selling devices, increasing margins, and expanding product lines.

But the game is not over.

Apple has always been a “laggard” in the best way possible. They take new technologies, they do them late, and they perfect them. I think once they jump into this generative AI game—and it looks like they’re going to jump in big time this spring—adoption is going to explode.

The Power of Default

It isn’t just about the LLM experience; it’s about the platform advantage.

• Apple Devices: There are 2.35 billion active devices worldwide on Apple platforms.

• ChatGPT: Currently has about 800 million weekly users.

We are comparing apples to oranges here, but the scope is undeniable. One thing I’ve learned as a PM is that the Power of Default is incredibly strong. If you are the default offering on a device, it is really tough for competitors to beat you. When Siri gets its act together, it will be the default on nearly 3x the devices of its competitors.

The Gemini Rumors

Rumor has it that Apple is going to use Google’s Gemini as the core of their LLM offering.

If this is true, it’s huge. Gemini 3 is benchmarking better than GPT-4 in many areas. Right now, my pet peeve with Gemini is that I have to open a separate app. If I’m driving or running, I want to use the hot button on my phone.

If Apple integrates Gemini into the native Siri experience—hosted on Apple servers for privacy, with a handoff to Gemini for complex search queries—it’s going to be a massive win. You get the engine of Gemini, but within the comfort and familiarity of the Apple ecosystem.

Come March, I think usage is going to skyrocket. Sam Altman is right to view Apple as a main competitor.

📉 AI Job Market: A Reality Check

I’ve been thinking about the “AI taking our jobs” narrative, and my perspective is shifting.

I recently read an article suggesting that LLMs might not be the “final form” of AI architecture, but rather a stepping stone. More importantly, I think we are currently in the “Mainframe Era” of AI models.

The Mainframe Theory: Right now, we have centralized models. We send a request to the “mainframe” (OpenAI, Google), it generates an output, and sends it back. We are likely heading toward the “Personal Computer Era” of AI, where everyone runs their own models connected via the internet.

We are still in the early innings. I used to think the job loss waves would come fast. I don’t really believe that anymore. Why? Because I’m swamped.

I’m working on AI products, and I have more work now than ever. Our engineers are still working hard to bring products to life. The way we work is changing—similar to when Excel rolled out—but the jobs aren’t vanishing overnight. Old tasks will fade, new ones will take their place, and the trick will be riding that wave rather than drowning in it.

🖨️ 3D Printing Hell

You’ll notice I don’t have a specific robotics update this week. That is because my 3D printer is currently my enemy.

The situation: My printer is jammed. I turned it off because I was genuinely pissed at it.

I feel like the guy in the stock photos with his head in his hands. Basically, I tried to “Emergency Stop” the machine. Pro tip: Don’t do that. I should have let it stop normally. Instead, it threw error after error.

I have a Creality K2 Plus. When it works, the parts are beautiful. But right now?

• I’m finding filament in places filament should not be.

• I had to replace the nozzle.

• There are multi-spool disasters.

I am slowly turning into a 3D printer technician, which is not what I signed up for. It’s been three days of troubleshooting. If I can’t figure it out soon, I’m calling a professional.

🏈 Quick Sports Update

I was bummed to see my Texas Longhorns miss the playoffs. Maybe we won’t play Ohio State next year (or whoever the equivalent is) and we’ll be 10-2 and in the playoffs. But, Hook ‘em. It was a great season, and I’m excited for the bowl game.

Hope this was informative and thought-provoking. See y’all next week!

Hope you enjoy this episode!

As an aged father of little ones, I can’t write AND podcast at the same time. My sleep-deprived brain just can’t handle it.

So yes, the written version below is, in fact, AI generated based on what I discuss on the pod. Thus, the best experience I think is the audio, but in case you can’t experience that, a written version is below.

Let me know if you enjoy it by subscribing below, I hope you do! See you next week…

Subscribe now

Slides are also here for visual reference:

https://gamma.app/docs/Game-Day-Gemini-Thinking-and-Building-R2D2-Mini-iuwljzzmor6wu2o

Robotics Scholar: Week 48 Recap

Gemini 3 is brilliant but slow, the “Code Red” at OpenAI is real, and I’m pivoting the robot build.

Welcome to the Robotics Scholar Podcast (formerly the Millennial Podcast—we re-branded!). This is Week 48. If you love all things AI and robotics, you are in the right place.

We have a lot to cover, from Google’s ecosystem dominance to building R2D2. But first, I have to say it: Hook ‘em Horns.

I had a great Thanksgiving watching my Longhorns beat the crap out of Texas A&M. Big Brother is back, baby. If you’re an Aggie fan, I’m sorry, but we are the best football team in Texas. It was a lot of fun, and fingers crossed we make a run in the College Football Playoff.

Now, let’s hop into the AI stuff.

The Gemini 3 Dilemma: Brilliant, but Slow

I’ve officially switched to using Gemini over ChatGPT, especially given the new Gemini 3 upgrade. As I mentioned last week, Gemini has huge advantages simply by being part of the Google platform.

However, I have one major gripe: It is so slow.

It’s amazing to use Gemini 3 with personalization, memory, and reasoning. You get incredibly high-quality answers. But man, the speed is an issue. I don’t think they could do “Gemini 3 Fast” fast enough (pardon the pun).

I truly believe that once they fix the speed, their usage and the quality of the experience will skyrocket. I keep looking it up, asking, “When is Gemini 3 Fast coming?” and the rumors say end of the year or early next year. Please, Google, give it to me. I’m on YouTube now asking: Give me Fast.

The ChatGPT “Code Red” is Real

We need to talk about the “Code Red” at OpenAI because it is very real.

The reason Gemini is winning right now isn’t just the model; it’s the experience surrounding the model.

For example, I wanted to put all my notes into ChatGPT to drive my planning and productivity experience. I simply can’t do that. It’s buggy. Canvas doesn’t really work for that purpose yet.

Meanwhile, Google has Google Calendar, Google Tasks, and Gmail all integrated natively. Everything just kind of works. This is where ChatGPT is struggling to compete. I’m even thinking about moving to Antigravity for coding.

I think Sam Altman is right to be worried. This is going to be their biggest challenge over the next few months. Competitors are coming. They need to cancel those trillion-dollar data center plans, put their heads down, and fix the product experience.

Robotics Update: The Pivot to R2D2 Mini

Since changing the personality tones for Gemini, the AI has been very forthcoming with me. It’s been “telling it like it is.”

My original plan was to print a life-sized R2D2. I thought it would be a fun way to learn robotics and how to use the Orin Nano chip I have. But the reality? That project is going to take 6 to 8 months to print.

Gemini basically told me: “You’re wasting your time building life-sized versions. You should think about using R2D2 to learn, but focus on designing your own robot.”

So, I listened. We are pivoting to the R2D2 Mini.

It’s a great learning device. It’s simpler—just three wheels, navigation, stopping, and those classic beeps. I suspect the LLM was trained on Star Wars data from the 70s and 80s, so it should handle this well.

I’ve already got the base here and just finished printing the right foot. We’re going to do this for the next month or two to get the fundamentals down.

The Big Challenge: Designing From Scratch

This brings me to the second part of Gemini’s advice: How do you design your own robot from scratch?

I have been in software engineering, design, and product management for over 15 years. But taking a physical hardware product to production? That is a massive new challenge. Honestly, I don’t know how to do it yet.

Early research points to using platforms like Fiverr or CAD Corner for sourcing designs. I don’t know if that’s going to work, but it’s going to be an adventure finding out.

The MVP Concept: I want to keep it simple. I’m thinking of a cute home monitoring robot.

• It checks if the stove is on or off.

• It helps ensure you unplugged the curling iron.

• It monitors rooms while you are away.

Nothing crazy. Just a practical, fun experience with a cute robot face. Eventually, maybe it evolves into doing laundry, but for now, we focus on the form factor and the basics.

Wrapping Up

We’ve got the tools. We’ve got the Jetson. We’ve got the 3D printer. We’re going to figure this out.

I hope you guys enjoyed the rebrand—calling this the Robotics Scholar feels right (the “Millennial” thing was dating me a bit). And I hope you enjoyed the look at the new “studio” (aka the playroom with the kids’ Elf on the Shelf watching us).

I’m plugging away and hoping to get something out in December or January. See y’all next week!

Thanks for reading/watching. If you have any tips on hardware design or CAD resources, drop them in the comments!

As an aged father of little ones, I can’t write AND podcast at the same time. My sleep-deprived brain just can’t handle it.

So yes, the written version below is, in fact, AI generated based on what I discuss on the pod. Thus, the best experience I think is the audio, but in case you can’t experience that, a written version is below.

Let me know if you enjoy it by subscribing below, I hope you do! See you next week…

Subscribe now

Welcome to week 47 of the Millennial Product Manager podcast—I think that’s what we’re calling it now.

I am back down here in my dungeon. The 3D printer is going here in the kids’ playroom, and we are preparing for Thanksgiving. It’s going to be great. I won’t be doing a pod next week, obviously; I think everybody’s going to be in their food coma, drinking, watching football, and hanging out with loved ones.

I grew up in Dallas, so I’m taking the whole family back there. We get to watch the Texas game, and... I’m worried. As a Longhorn alum, I am worried. But we’ll see how this week goes.

It has been a big, eventful week. Let’s get into it.

Gemini 3.0 and the Platform Advantage

The first thing I wanted to talk about is obviously the huge release from Google this week with Gemini 3.0. I’ve been using it, of course. It’s impressive. The “pro” part—the thinking part—is a bit annoying; hopefully, they release a version of Gemini 3.0 Fast that I can use for day-to-day stuff.

But what is happening as a result of this release is that I’m really reconsidering my tech stack here. I used to think I was going to be a ChatGPT user going forward, and that was just the platform I was going to use. But honestly, the Gemini 3.0 release has really got me thinking.

This, combined with the news that Apple is likely going to use Gemini for their Siri implementation in March, highlights something important. When you use something like Google or your Apple products—your iPhone, iPad, MacBook—those offerings have a robust platform that comes with the AI solution.

It’s not just about Gemini. It’s not going to be just about whatever Siri looks like in March. It’s the fact that all that stuff works well with Gemini.

One of the things I’ve noticed is that Gemini can search all of my Gmail. I can ask questions about Gmail. It works natively. It can pull stuff from my calendar. It works with Google Search, which is obviously a powerhouse and incredibly accurate. They’re not building stuff from scratch. When I use Siri, it works with my Apple Reminders, Notes, and Mail. It has all of my previous account history in both places.

I think going forward, the winners in the AI space are going to be the ones that couple AI with their platform advantage.

It’s not just about the models. The models are great—having a frontier model is important—but it’s everything around that experience. It’s the full offering that gives something like Gemini an advantage.

Don’t discount the platform investment. Don’t discount the platform experience that you’re going to offer along with the AI experience. What we’re finding is that AI by itself is not enough.

The Struggle to Find Value in Software

If you work in tech, or if you work the kind of job that I do as a product manager, there’s always this itch to be entrepreneurial and do something on your own.

But one of the things I’ve noticed is that I’m really struggling with what I would want to invest in on the side, or as a little startup in software. I actually am struggling with just what is even valuable anymore.

Like we were talking about earlier: Platform plus AI is really valuable. It’s not enough to go build an AI-based chat experience for X, Y, Z thing. That’s not enough. I’m finding I’m having a hard time thinking of things to make that leverage AI but are valuable without the platform investment that’s required. And creating platforms is hard. Creating marketplaces is hard. It’s a lot of grind.

It changes every hour. What’s actually valuable? What’s actually useful? Whole startups are getting destroyed, whole companies are getting destroyed overnight.

There’s a big rush, and people are taking chances trying to figure out what’s going to be worth it, but man, it’s really difficult to figure out what to invest in on the software side right now.

Elon Musk and the Sunk Cost Fallacy

I came across this video of Elon Musk talking about the five key things he’s learned building things, standing in front of a rocket. I think a lot of us have seen this video.

One of the things he talks about is how often engineers and technical people don’t ask the question of if the thing that they’re working on is the right thing to be working on. They’re so heads down and focused on optimizing the thing that exists already, that they don’t stop and ask the question: Why? Why does this product exist? Why does this experience exist? Why should it continue to exist?

I’ve actually been running into that a lot lately.

Coming back to the point about how AI is totally changing the game on what’s valuable in software—there are things that I have built, or that have been built as part of my job, that I actually doubt are valuable anymore. I doubt they should continue to exist in this new world that we’re all living in.

But I’m finding that it is actually difficult to get people to move on from these implementations. Even things that were built six months ago or a year ago. It reminds me of that point Elon is making: sometimes due to sunk cost fallacy, it’s hard to walk away.

In this world we’re living in with AI, you have to ask that question more and more. Just because a system exists doesn’t mean we shouldn’t build something completely new.

Link to video:

The Steep Learning Curve of Robotics

Finally, as you know, I’ve been on this robotics adventure. You can probably hear the 3D printer going in the background.

The learning curve for robotics is steep, and that includes things like 3D printing. It’s taken me a lot of test prints to figure out how to properly print things. If you follow me here on Substack or on Instagram, you’ve probably seen some of the examples, like R2-D2.

I printed a large version of him with a ton of supports, but the problem was those supports were just glued onto him. I couldn’t get them off. It was impossible.

Then I learned how to use tree supports, and that’s gone a lot better, but I still have to learn how to remove them properly.

Then I had to transition software. The Creality printer I have comes with a default software that was just messing everything up. So I had to switch to Orca Slicer and get that dialed in—learning from TikTok and YouTube how to properly print things on Orca Slicer.

Now I’m messing with this Jetson Orin Nano board from NVIDIA, and I’m back in the engineering days of flashing firmware, loading stuff via an SD card, and installing an SSD to expand memory so it can handle real-time models and computing at the edge.

It is just a lot to learn.

Honestly, though, it’s probably the most fun I’ve had with technology in a good while. The more time I spend with it, the more I’m like, “Oh yeah, this is fun.” It’s hardcore engineering stuff that I learned in school. It fulfills a need, especially given the flux I’m feeling in the software world regarding what is valuable to build.

It is steep. It is not for the faint of heart. But few things that are worth doing are easy.

Something you think about when you’re 41 is that you think you’re done learning—that you’ve learned everything you can and you can’t learn quickly anymore. It turns out you’ve still got a pretty good motor. I definitely feel that way as part of this journey. It’s challenging when you have kids and your schedule looks different, but man, is it fun.

That’s it for this week. Thanksgiving is around the corner. Today, it felt like collectively everybody at the company just said, “Alright, Thanksgiving is almost here, let’s give up,” and canceled a bunch of meetings. I think that’s great. We’ve all worked really hard.

Happy Thanksgiving, everyone. Hope you enjoy the break, and we’ll see you when we get back.

Cheers!

As an aged father of little ones, I can’t write AND podcast at the same time. My sleep-deprived brain just can’t handle it.

So yes, the written version below is, in fact, AI generated based on what I discuss on the pod. Thus, the best experience I think is the audio, but in case you can’t experience that, a written version is below.

Let me know if you enjoy it by subscribing below, I hope you do! See you next week…

Subscribe now

📍 Working From the Kids’ Playroom (And Somehow Liking It)

Good evening, everybody. It’s been a long time since I’ve done one of these, and honestly, this whole stream-of-consciousness podcasting thing might actually be easier than trying to write a full newsletter every week. You have no idea how long those take. Even the simple ones. They look like 15 minutes of typing but they’re not. They’re hours.

So yes — I’ve officially been banished to my kids’ playroom downstairs. That’s my office now. Oddly… I kind of love it? They drift in, grab toys, run back out. I see them more. It feels… nice. Very “dad grinding in a bunker” energy.

If you hear a machine humming in the background on the pod, that is my new 3D printer doing its thing.

🤖 Robotics Season Begins

This leads me to the first topic: robotics.

I’m really going to give this a shot. What’s happening in robotics right now — Figure, Neo, Apptronik, all of these companies dropping new robots — it’s wild, and kind of inspiring. It made me want to dust off the robotics part of my engineering degree from 20 years ago. I cannot believe it’s been 20 years. That hurt to say out loud.

So yeah, I’ve got a Jetson Orin Nano Super sitting here. I’ve got this giant 3D printer printing calibration tests. I haven’t even gotten the board fully running yet. I’m literally crawling before walking. But I’m excited.

And then my wife said something that honestly stuck with me. We were watching the new Fantastic Four movie and she saw Herbie and was like: “Why can’t robots just be cute? Why does everything have to be this sleek creepy humanoid thing?” And I was like… yeah. Why can’t robots just be cute? Why can’t they make little beeps and boops and roll around the house like Star Wars droids?

So maybe that’s the idea: make cute robots. Friendly robots. Household robots that don’t feel like they’re about to file a report on you. I like that direction.

I don’t know if I’ll make anything good. I don’t know if I’ll get beyond testing the printer and booting the board. But I want to try.

🧠 Context Engineering: The Real Work Behind AI Products

On the work side, I’ve been spending a lot of time on context engineering. And here’s the thing most people don’t realize when you build AI products: the models are incredibly capable. They’re powerful. The hard part isn’t usually the model. It’s everything around it. Especially the context.

If you give a model the right context, it does great. If you give it the wrong context… chaos. Hallucinations. Nonsense. So getting the right stuff to the model, in a format it can read and reason over, is the real job.

A lot of the systems we work with are still built for the “old world” — deterministic systems, clean inputs, clean outputs. They weren’t built for non-deterministic systems reading PDFs, documents, contradictory information, structured data, images, all of it. A lot of “normal” search chunking doesn’t work well for LLMs because it strips out too much context. The LLM actually needs the big block of text. The surrounding paragraphs. The page. Sometimes the whole document.

So recently we’ve been expanding our chunks, sending more human-readable info, and pushing more context because these models can handle it now. Larger windows, better reasoning, multimodality… it’s honestly amazing how much more you can send compared to two years ago.

And the thing that helps the most? End-to-end traces. Seeing exactly what happens in the system. It’s the closest thing to debugging an LLM product that feels like real engineering. That and experimenting. Lots of experimenting.

💪 Health Journey: 208 → 190 and Still Going STRONG.COM

Okay — last topic on this little test run before I go to sleep: health.

I started this journey at 208 pounds. I’m at 190 now. Doctor says I need to lose another 20. So I’m maybe halfway. But it feels good. I can actually see the difference in how I move, how I run, how my back feels.

Here’s the big thing I learned: calories matter more than you think. Especially when eating out. We basically stopped eating out because I realized I have no idea what’s in any of those dishes. That chef video of the guy making scrambled eggs with a stick of butter was burned into my brain. That’s why we’re all overweight. It’s not even our fault. No one knows what’s in anything.

Walking helps a lot. Getting lighter helped me run again, which I honestly thought was over for me. I have a bulging disc so I figured, well, that’s that. Turns out losing 10–15 pounds relieves… a lot. And now I can run. And that improved my VO2 max from 28 to 40. I had no idea that was possible at my age.

The weird thing is my Oura ring recently started telling me my cardiovascular age was like six years older than I am. And I was like: dude… what? I’m running! I’m working out! I’m eating better! But then I realized weekends aren’t rest. They’re kid-wrangling. Which is arguably harder than running.

So now my rest days are Tuesdays and Thursdays. Actual rest. And my numbers are improving.

The next 15–20 pounds are going to be brutal. But I’m excited. I feel good. I feel like I’m on the right path.

🎤 Wrapping Up

Anyway. That’s it. End of podcast.
Thanks for dealing with the 3D printer noise — the thing is a monster.
I’ll post a picture of my robotics workstation in the written version.
I’m excited for this next chapter. Maybe the podcast continues weekly. We’ll see.

See you next time!

I’m using this newsletter to share at least three things l’ve come across this week to help you build better product. Subscribe so you don’t miss these when they come out:

Subscribe now

On to this week’s thoughts and updates:

1. Focusing on your customer is always a good answer

I was on a customer call this week. We started our conversation with a discussion about the features they’re looking for in our products. Which can be a bit nerve-wracking, you do a lot of work as a product manager to make sure these conversations go well and you’re on top of what your customers need.

You can imagine how rewarding it is then when, on this call, you hear a feature that you just released that they want. “Can you support Spanish as a language to chat with?” Of course we can Ms. Customer! Of course we can.

Why were we on top of it with this customer? We had heard from other customers that Spanish was important for their users. We had done our research that showed that supporting Spanish would open up new markets and swaths of users. We did our planning, leveraged our mountain of evidence and made the right call to support it now to meet our customers needs.

We focused on our customers’ needs and it paid off. It usually does.

Does it mean doing everything that your customers ask? Not necessarily. Not everything makes sense, many things are customer specific for example and just aren’t high opportunity.

But more often than not your customers are great at communicating what they need. You just have to get really good at listening and responding.

2. We Innovators always think we’re further on the adoption S curve (we def aren’t yet for AI)

I’ve been making products for a little over 15 years now. I’ve seen a lot of shit go up, down and sideways, but there are still a few constants.

One constant that shows up over and over again is the product/technology adoption S curve.

On this S curve, something I see a lot is that those of us working in tech that live and breathe this stuff every day are on the front of this curve. But most people are in the last two fifths of the curve. And it’s pretty consistently true, since I’ve been alive we’ve had the PC revolution, the internet, mobile, crypto and AI all truck through here. And almost every time, as someone in the Innovators section of the curve, I encounter so many people that believe deep in their bones that we’re farther along the curve than we actually are.

Myself included! Early in my career, I bet on myself and did a startup trying to connect energy efficiency data together in one unified dashboard like Mint.com if you’re old enough to remember that. 😭 That reality still hasn’t materialized…15 YEARS LATER. All of us innovators think we’re later on the curve than we actually are. It’s our burden to bear.

So when I came across the article below, I wish I could say I was surprised but I wasn’t:

AI Supremacy

How do normal people view Generative AI technology?

If you think this topic is important, why not share it with a colleague, friend, or acquaintance. It’s a revealing survey to say the least…

Read more

7 months ago · 74 likes · 1 comment · Michael Spencer

Most people still use ChatGPT the way they use Google, mostly keyword search. And in the article, they talk about how most people can see AI take on simpler things like forecasting the weather, but not more complex and human-nuanced things like counseling on God and faith (spoiler alert: AI can do that too!)

We still have a long way to go up the S curve. I suspect it will happen somewhat faster than with other technologies, but we’re still in early innings.

3. Recommended Reading: The Secret of Secrets

Okay okay, I know I usually share something useful in terms of books. But dang it all, this book is super fun and great listening while taking breaks or walks. I’m a sucker for Dan Brown novels, so this one doesn’t disappoint. And as a sign of the times, this book has already been picked up for a Netflix series. I’m 70% of the way through, can’t wait to see how it ends! Probably this weekend 😂

That’s it for this week!

If you’ve found this helpful, please consider showing your support by subscribing:

Subscribe now

I’ll be back around this time next week with more useful product manager things!

WE’RE BACK! Happy Friday everyone! Welcome back to The Millennial Product Manager, your weekly product manager newsletter.

I'm using this newsletter to share at least three things l've come across this week to help you build better product. Subscribe so you don't miss these when they come out:

Subscribe now

On to this week's thoughts and updates:

1. FINALLY hitting a stride

Look gang, starting a new job is just hard. You have to learn about the people you work with, the technology and product you’re working on, the industry you’re working in, the current state of things…it’s not easy. It takes time to get a solid footing.

In my case, the last six months I’ve been getting onto the on-ramp of this new crazy world we find ourselves in with AI. I went from being a Product Director to essentially becoming an IC Technical PM again. And on top of all that, I had to learn about the construction industry. And I’m 41 with a family! My brain isn’t what it used to be lol. Not a small feat, but here we are six months into it and I finally feel like I can hold my own.

If you’re on this journey too, just know that every day and every week things get better and better. You’ll be the expert in the room again before you know it!

Speaking of learning more about building AI products…

2. Stable, clean eval data is crazy important

Evals are super important for judging how your AI-based product works. The underlying model you’re using is trained and tuned by the model provider of course. But your job is essentially to teach the model how to meet the needs of your customer for your specific use cases.

“Evals” or evaluations evaluate how your application performs once you’ve given the underlying model instructions via the prompt and context via selected documents or endpoints like a search endpoint for example. It’s a lot like giving someone a test, and you pass and fail them based on how well they answer the questions on the test.

What I’ve learned lately is that it’s important to have a stable, clean dataset to test against. Why? If we use the test analogy, imagine administering a test but the answers keep changing. It’s going to be hard to truly evaluate that person’s understanding of the material right? Right.

Stable, clean data sets for evals keep your answers consistent and predictable. You know how the answers are shaped so you can properly evaluate the product experience. And, you can change the eval, or test in the example, when you know for sure that the underlying data set has changed.

So, if you’re building a product, make 100% sure you have a stable, clean eval set! One problem solved for your product, on to the next problem: adoption.

3. Adoption still doesn’t solve itself

I’ve had a few meetings in my career now where someone says something like “oh adoption is easy, just slap an in-product notification up there and we’re good! If the product is good users will just show up!”

Nothing could be further from the truth. This ain’t Field of Dreams yo!

AI is amazing, but there is no shortcut to solving adoption problems. YOUR PRODUCT HAS TO HELP DRIVE ADOPTION. That has been true since I started as a product manager 15 years ago, it is true today, and it will be true on into the future until AI become sentient, all-knowing and all-powerful. But by then we’ll be dead so it won’t be your problem anyway! lol

Seriously though, if you need adoption of your product…get it on your roadmap. Dedicate product energy towards it. Otherwise it’s not happening.

4. Recommended reading: AI 2027 (there’s a podcast version!)

Working in the AI field, I have to admit that I am simultaneously amazed and worried by the breakneck speed of progress with AI. It seems like every month we’re hurtling closer and closer to many unknowns related to our world with AI in it.

Which is why I found AI 2027 to a thought-provoking exploration into those unknowns with the lens of only a few years out. It’s hard to say what’s going to happen beyond 2027, so I’m appreciative of how this content was approached.

I realize this has been out for a while now, and in true dad-of-young-kids fashion I *just* got around to reading it lol but if you haven’t yet I’d recommend giving it a read (finally!) And if you’re like me and haven’t been able to sit down and read a book in 6 years, I highly recommend the podcast version of it.

It’s not as rich as the site version of course but I was much more likely to finally read it so there!

That's it for this week!

If you've found this helpful, please consider showing your support by subscribing:

Subscribe now

I’ll be back around this time next week with more useful product manager things!

Today we’re going to talk about something that’s been on my mind for a long time. I’ve been a corporate warrior for almost 15 years now. And the entire time I’ve been in corporate, I’ve hated taking notes in a meeting.

My brain just can’t do it. I can’t engage and type notes about what’s happening at the same time. I’d argue that, biologically, none of us really can. We’re not meant to multitask clearly from the research.

But STILL, in the year of 2025 I see people click-clacking away taking notes for everyone in a meeting. WHY?!?!?!?!? I think it's a total waste of effort and time to be manually transcribing notes in a meeting now. Here’s why, and I’d be willing to bet that in a few years taking notes will be a relic of the corporate past. Good riddance.

LLMs Transcribe Conversations Better Than Any Human

First let’s talk about LLMs. LLMs and machines can take notes for us much better than a human ever could. An LLM like OpenAI GPT-4o with Advanced Voice Mode definitely can understand you now. Their voice transcription is excellent with 99.9% accuracy. And it’s always on, never not paying attention if you want it to. Humans by comparison with our little monkey brains are totally fallible. LLMs are much better at this job, so why wouldn't you just let an LLM voice transcriber do that job? Combine this with the next technology and it really makes no sense why you’re typing notes at all…

Cloud Tools Auto-Organize Your Notes (No Copy-Paste Needed)

Cloud-based software has now made it so that an LLM can record + auto-save all of these perfect transcriptions, AND write a summary in whatever format you want, automatically. You literally copy them, paste them into Slack, add links if needed and 💩 poop them out to the required Slack channel. So again there's NO REASON to be manually writing notes down, machines have us beat. But the worst thing of all to me is…

Typing or Handwriting? You’re Guaranteed to Miss Stuff

Let's just say you're taking notes in like a Google Doc or notion with your hands typing like this little guy, who looks like everyone I’ve ever seen taking notes in a meeting:

Be honest now…

Are you really 100% paying attention while you’re doing this?

Are you actively engaging in the discussion to make a critical decision?

I DOUBT IT.

We’re all little monkey-brained humans parading around as if we’re omniscient, omnipresent machines. It’s a lie! You're bound to miss stuff while you’re typing frantically. It does no one any good.

It was understandable in 2015. We lived in the total stone age it seems. But LLMs + cloud-based storage already do this with much higher accuracy than a human can. And, while we’re being honest…

You're not likely taking notes to remember stuff for yourself. You're likely taking notes to share them with people afterwards. Critical decisions, action items, the lifeblood of corporate tasks that keeps things humming. And we're also doing it because it's the expectation at a company that notes are taken a record is logged.

Which makes sense, but even for that important purpose we’re using the worst tool imaginable: HOOOOMANS. Machines are better at it.

And finally, while I’m on a roll…

Most People Never Re-Read Shared Notes Anyway

Aaaaaaand there it is. Half the time, people don't even read the notes people stumble over themselves to write.

If you were in the meeting yourself, often you remember your action items you're referring to in your own notes. You might use the generated notes in case you forgot something or that you need to review your action items in the notes to make sure that those action items match up with the action items that you've assigned to yourself. Excellent use case.

But I’d argue 80% of the time, you’ve got your own notes and you’ll never look at the created notes again. 🙃

AND, if you're relying on a human that's fallible to make those types of notes for you, then you've already lost. Humans can't remember all that stuff, especially if you're trying to engage in the meeting and take notes at the same time.

So it's all a waste of time. Hopefully I’ve made my point now, and if you’re not using a tool to take meeting notes for you, you’re wondering what are some good ones. I think the best tools are the ones you have access to…

Granola, Notion AI, Zoom Meeting Notes—Whatever Gets It Done

I’d recommend checking any of these out that fit your corporate policy:

1. Notion AI. This is the one I typically use. It’s harder to get started compared to Granola but Notion/Claude say they don’t train on your data. Big plus for corporate meetings.

2. Granola. I love their UX and experience, but it doesn’t seem like I can guarantee non-training. But it’s secure otherwise and an amazing product so I’d recommend.

3. Zoom Meeting Notes. This one requires recording the meeting, which it seems like many teams don’t love. I agree, we really just need notes. But if this is the only option you have, it’s a good one.

Use any one of these. They spit out great meeting notes, complete with action items and key decisions you likely would have missed typing away. Save yourself.

Still Scribbling? Tell Me Why You’re Lighting Time on Fire

So dear reader, as you can see I believe strongly that if you’re taking meeting notes manually in the year of our lord 2025, you might as well take your time and just light it on fire. Personal notes? No issues here. I take them all the time. But meeting notes? Manually?! Ridiculous. That being said…

If you’re still doing it and plan on continuing…why? I would love to learn why. Comment below!

The stuff that’s done the best for me on TikTok has been mostly entertainment related:

@millennialpmIt’s hard work but it’s honest work 😅 #productmanager #developer #engineering #uiux #uiuxdesign #techtok

Tiktok failed to load.

Enable 3rd party cookies or use another browser

Thus the hypothesis made sense. But shooting these things means getting all cleaned up, turning on lighting, getting the mic fired up…it’s a lot of work! And it’s hard to get the look you’re going for.

So I had a brilliant idea: let’s make videos with AI! Sora creates amazing video, it can just crank this out for me. Well, not really. Here’s a sample of what it gave me. 👎 Gross.

So then I had ANOTHER brilliant idea: let’s use ChatGPT image generator and Runway to animate these beautiful images! The results so far have been mixed. The images coming back are great:

But they’re not coming back super accurate to what I look like. Nor are they animating 100% well when I stitch them together in Runway. Here’s a piece of what I had in mind:

Anyway, making AI videos is hard. I’m going to keep giving it a go, will keep y’all posted. 😅 Don’t forget to subscribe if you’d like to keep up with the latest:

Subscribe now

Have a great Friday!

Subscribe now

I get asked pretty often which AI tools I’m actually using these days. Probably because there are a lot of AI “fart apps” out there, i.e. apps just using AI for marketing/getting funding/to be cool but their approach is not actually useful.

I’m not going to name names 👀 instead I think it’d be more useful and frankly, more fun 🤩 to talk about the AI products that I not only absolutely love, but use for actually useful stuff everyday.

One more thing before we get into the goods…this is about NEW products. Products that were not really here without AI and LLMs specifically. I’m not including things like Google Photos, Spotify, etc. that always have been and continue to be amazing. Still using them, not alternatives YET. 👀

So here we go, as of May 2025, here are the products I’m using daily at this point:

ChatGPT: My creativity partner

DUH NO SHIT AJAY, so is everyone. Over 400 million users agree that ChatGPT is the bomb.com, it’s true.

Something interesting though is that most people still use ChatGPT as a search engine mostly. ChatGPT has made this use case better with SearchGPT. But I honestly don’t think it’s the best for that. It gets really buggy and isn’t as rich as our next product.

The features I really love with ChatGPT though are:

1. Image generation. HOLY MOLY, there’s really nothing like it.

2. Advanced Voice Mode. I have yet to find a better AI voice assistant. I will talk to it while I’m driving or on walks sometimes. It’s really incredible.

3. Brainstorming writing. Honestly, I’ve found the writing output to be unusable for things like this. But I DO use it for brainstorming and to check if there are any gaps in how I’m planning on talking about something. Great for that.

The use cases I find myself using it for the most are:

1. Generating images for short films I’m trying out or just for fun from family pictures

2. Brainstorming creative writing for this newsletter, emails, texts if they’re going to be long and detailed

3. I’ve used to help me search big documents and images. Pop in the PDF, and you can ask it questions about the PDF LLM style.

4. Advanced Voice Mode for job interviewing, one of my favorite use cases. Give it a Linkedin profile of the person interviewing you, along with any notes. You can mock interview and practice responding to a live person, along with feedback at the end on how to improve. And you can do it whenever you want. Amazing.

Now let’s talk about why I don’t use ChatGPT for search.

Perplexity: Epic for research, search

The undisputed champion of search in the AI era is Perplexity.

First of all, the design is gorgeous. Minimal but modern and just well-thought in all the best ways. They did hire one of my favorite designers to work with, so I’m not surprised. 😃

What keeps me coming back though is:

1. I don’t get ads, and for a subscription price that’s totally worth it I can search agentically as much as I want. I hate ads.

2. On Google, I have to go Search → Search → Search → Click → Back to search → Click before I’m happy. On Perplexity since it’s agentic, I go Search → AI agentic search does all of the wandering around → Click and I’m done. So much faster and less random wandering.

3. Up to date information by default. Sorry ChatGPT, but it’s stupid that I have to click on “Search” if I want the latest info. Perplexity does this by default. It’s 2025. I know they’re working on it, but ChatGPT should know I want the latest info by default.

Anyway, the use cases that Perplexity now has from me are:

1. PRODUCT RESEARCH, this is the killer app. I save hours with Research from Perplexity, getting it to help me research ideas for features I’m working on, improvements to the UX I’m thinking about that others have tried. It’s saved me hours of time wandering around the internet.

2. Pretty much any search that comes to mind goes to Perplexity at this point.

3. Sports scores, hallelujah!

4. Surfing the news, personalized for me without all the damn ads Google lets through

Perplexity changed my life. Speaking of life…

Cal AI: Shed fat the simple way

This is a recent change, but Cal AI is a pretty great nutrition app.

Oura Ring’s meal logging would take the cake 🍰 but I just can’t get any calorie tracking out of that. And unfortunately calorie tracking matters for my body for losing fat (oh how I wish it didn’t 😅)

Cal AI mashes together the photo logging + GPT-4o image features with calorie and macro tracking which is just easy. I take a photo, GPT-4o image analysis does a great job. I tested this out on ChatGPT and it was great but there was nowhere to store it before…now there is! Thanks Cal AI.

Someday I hope to not do meal logging anymore. But for now, Cal AI makes it pretty seamless. Now let’s talk about something completely different!

Runway: Testing out creative video with AI

So I tried Sora for a fun idea I had for a video and the level of slop I got out was astonishing:

@millennialpmGuess we have some work to do 😅#severance #sora #ai

Tiktok failed to load.

Enable 3rd party cookies or use another browser

I started poking around the internet for ideas on how I could do what I wanted to do and came across this overview on Sora with Kling/Runway:

Kling was used for that viral Studio Ghibli LOTR video that made the rounds. I liked Runway a bit more, but the general gist is you can make short films with AI and it’s magic.

I’m trying to work out a short film with it, so far it’s incredible. I’m looking forward to sharing what comes out of it!

Lovable: I LOVE it for prototyping

This product needs no introduction. If you haven’t been living under a rock, you’ve heard of the startup rocket ship that is Lovable.

Cursor and OpenAI’s new baby Windsurf are amazing, no doubt. But it’s just so fun to fire up an IDE in your browser and make cool shit. I talked about this last week in the newsletter, it took 20 minutes and a few prompts to create a new site on Lovable. It’s just the best and so smooth. Anytime I have an idea about something I want to do, I use Lovable by default. It’s just fun.

Granola: Actually pay attention during calls

Coming back to the list of things that save me hours of time, one of the best things on that list is Granola. OMG.

What a LIFESAVER. It listens to your system audio (with your permission of course) transcribes on your machine, and then runs it through an LLM to create perfectly crafted notes so you don’t have to type jack shit during your meetings.

Who the f is taking notes during their meetings anymore?? WHY?! Back when I was managing people I was like yo don’t bother, just download Granola and use that. Waste of time to take notes for real. Granola is the best.

Matic Robots: The robotics revolution is here

Listen, I didn’t plan to get excited about a robot vacuum. My iRobot was fine, even though I almost never used it and had to turn it off because it was loud and annoying about getting stuck all the time. 😅 Well folks, the AI revolution has given us a miracle in Matic Robotics.

This thing literally mapped my whole house with color imagery, identified all of the rooms, the materials on the floor 🤯🤯🤯 and cleans the house in a snap, missing toys and people and dogs. It’s incredible, I’ve never seen anything like it, and it’s probably not possible without next generation AI. I watch it tackle my house in awe.

Honorable Mentions

Claude: Testing it out for work

Claude seems to slowly be taking over for work tasks. I had heard it’s better for work, in particular writing and coding tasks.

I’ve been wanting to try it out but just couldn’t figure out a great use case for it until recently I noticed that it doesn’t train on your data like Gemini.

Gemini can do less than Claude can, which is why I’m test driving it. But I do think this is going to be an issue for OpenAI going forward. If they continue to train on data, people are going to be less inclined to use it for work.

Anyway, giving it a go, will report back on how it goes!

Oura Ring: Advisor insights are getting good…

I love my Oura Ring. I check the app every day, multiple times a day. It’s for SURE in the category of products that weren’t enabled by LLMs, but inherently was best in class at ML and data-driven experiences.

They recently shipped Advisor. I was a beta tester for it, it was pretty rudimentary when I started with it. But lately it’s been giving me some banger insights and I am here for it.

One of the ones I got recently is the decline in my resting heart rate push notification as a result of more sauna time, walks and Zone 2 training. The notification caught me by surprise and was fantastic. Definitely a product I’m paying more attention to (as if I wasn’t in their app every hour already 🤣)

Reflect: My Notes Daily Driver, iterating on an AI first strategy

AI for notes is not a new concept, but Reflect is really infusing OpenAI and Anthropic into their notes app in a way that’s unique compared to Notion and others and I am here for it.

It’s native to the app, which is much smoother than Apple Notes. And I find the functionality less complicated than Notion, along with task management just being easier to do. Highly recommend if you’re looking for a Notes daily driver with AI.

Did I miss any new AI tools that you love? Leave me a comment. I’m always looking for great stuff! And don’t forget to subscribe if you enjoyed this and would like to read more:

Subscribe now

Hope you got some new stuff to try out of this list! 🚀

I'm using this newsletter to share at least three things l've come across this week to help you build better product. Subscribe so you don't miss these when they come out:

Subscribe now

On to this week's thoughts and updates:

1. The cylinders are finally firing a bit again, time heals many onboarding wounds

We’re not machines unfortunately. That’s what I have to tell myself anyway every time I get frustrated with the speed at which I’m able to onboard. The brain needs to connect ideas, and it just takes time.

But FINALLY, this week I’m not feeling completely useless. Not all cylinders are firing yet, but I’d say one or two are.

It’s nice to be useful again. I’m sure I’ll get knocked back down to earth at some point. But I’m going to enjoy the moment after transitioning to a new job.

Speaking of getting a new job…

2. If you need help landing that PM job, resources below

Let’s face it, the job market is terrible. So I wrote this post on what I believe is the most important factor in you landing a great job as a Product Manager…the referral.

Millennial Product Manager

Land that Product Manager Job: How to Get a Great Referral

👋 Welcome to a 🔒 paid subscriber-only edition 🔒 of my newsletter. The Friday post will always be free 🎉 These monthly-ish deeper dives along with messaging yours truly with any questions you have come with a paid subscription…

Read more

a year ago · 1 like · Ajay Waghray

I spent extra time on this one this week because it’s just that important and I want anyone looking for their first or next opportunity to succeed. In that one is a step-by-step guide and resources, I hope they’re useful.

On top of THAT, you can book time to mock interview with yours truly here! And if you’re a subscriber, let me know and you’ll get a special deal. Let’s get you that job!

3. One of the more useful AI implementations I’ve seen: LILLI @ McKinsey

Unfortunately in this frothy age of AI we’re living in, we’re going to see a lot of what I refer to as AI “fart apps”. Just lots of stuff that’s not that useful leveraging new tech just because. I already lived through it with mobile apps. The cycle continues. 🤷‍♂️

But with all of the froth comes genuinely useful things like Uber, Whatsapp, Doordash, etc. And here’s an example of a great use case:

So genius, this is one of those products that makes oodles of sense and has crazy product-market fit. 70% of employees using it?! Of course! Who wouldn’t want to have 100 years of knowledge, pitfalls that no one else has at their finger tips.

I feel this way about the copilot experience I’m building in the construction industry. So useful in cases where other tech just falls flat. I’m excited for this caliber of innovation.

Speaking of fart apps…

4. AI 2027 & a reasonable take on our AI future

I have to admit that I didn’t exactly know what AI 2027 even was until I ran across this excellent article first:

Understanding AI

Three great articles on the modest pace of AI progress

Last month, a team of prominent AI researchers and Internet writers published AI 2027, a website predicting that AI will dramatically change the world within the next three years. As long-time readers might expect, I did not find these predictions convincing. But rather than writing a full analysis of the AI 2027 scenario, I thought I’d highlight three …

Read more

a year ago · 8 likes · Timothy B. Lee

Mr. Lee does a great job of referencing articles that paint a more realistic picture of where we are with AI and market adoption.

Listen, I’m not going to be Mr. Curmudgeon here. I believe our future with AI is going to be very bright. But I’ve actively participated in one of these huge technology shifts in mobile before. One of the few things I know about big tech shifts like this is that:

1. The future is super murky. It’s extremely hard to know who/what will win/lose.

2. The future happens in fits and starts, which is exactly what’s happening with AI.

So maybe it’s 2027, maybe it’s 2032 or 2037. Maybe it’s OpenAI that wins, maybe it’s Google, maybe it’s DeepSeek, maybe it’s no one, maybe it’s someone we’ve never heard of. Maybe it’s agents, maybe it’s something else we haven’t even foreseen or seen yet.

I’m thrilled to be on the ride, I never thought I’d see this innovation in my lifetime. So in my view, it’s best to keep up as much as you can, stay alert for opportunities and just enjoy being part of it all playing out.

5. My Lovable love affair continues

I’m going to put together a video about this soon, but I wanted to see if I could rebuild my personal website with Lovable instead of paying Framer to host it and it did not disappoint.

The design quality output of Lovable and the way it leverages prior art to make you something great is fantastic. I gave it design examples from Dribbble and it just went to work.

I haven’t tried many of the other tools yet, but that’s in large part because of how much I love Lovable. And with Lovable 2.0 I can’t wait to see how that makes building on personal projects that much more fun.

More to come here soon, video is in progress…

That's it for this week!

If you've found this helpful, please consider showing your support by subscribing:

Subscribe now

I’ll be back around this time next week with more useful product manager things!

I know I know, paying for stuff isn’t fun. But it should be worth it if you get a 10% salary bump like I did with these tactics 💰 and you’ll get some resources to help you fast 🚀 let’s get this bag!

Story time!

I just started a new job this April. It took me 30 days to go from looking for a job in this pretty awful job market to signing an offer and receiving a 10% pay bump. And it all happened as a result of having one thing: a great referral.

Am I going to make the outrageous claim that you can do this too? HELL NAW. Maybe you can, but I’ve been at this whole product management thing for 14 years. Unless you’ve got a similar level of experience or more, chances are good it will take longer than 30 days.

But I know one thing for sure: I’ve never gotten a product manager job without a referral. Not one.

Why do referrals work?

I think it all comes down to human nature. Back when we were all cave-people, we survived and thrived by building trust amongst one another. This way, when the saber-toothed tiger came to eat everyone, we knew that we had each other’s backs.

We’re not too different from cave-people 10,000 years ago. We prefer to work with people we know and trust. And if we can’t do that, we prefer the next best thing…working with other people that the people we trust, trust also.

You might admonish me for being old-school. I don’t blame you. If you work in tech, at least at some point in your career you probably believed that tech is a true meritocracy, and only the most qualified will make it into the best jobs. In many cases that is very true, it happens.

And I’m not telling you that a referral will guarantee that you’ll get that job either. There is still much preparation, knowledge and skills that need to be demonstrated beyond the referral, no doubt.

But I want you to put yourself in the shoes of a hiring manager for a minute. I’ll help, I have been a hiring manager for product manager for 7 years. Let’s say you had two candidates:

Candidate A Did really well in his case study but had some less than perfect answers in the behavioral questions. He has no referral.

Candidate B: Did really well in her case study but had some less than perfect answers in the behavioral questions. She had a great referral from someone who has worked with her directly on a number of projects.

Who do you think I’m going to hire? I’m not sure if it’s obvious yet but DEFINITELY CANDIDATE B!! My butt is on the line, Candidate B I trust to help me bring home the bacon.

So a referral is not the only thing you need. But it can give you a significant edge throughout the hiring process.

What kinds of referrals work?

So here’s an important point. When I say referral, I’m NOT referring to an intro from someone you reached out to cold. I’m also NOT referring to someone you cold messaged on Linkedin.

Contrary to what many influencers say on TikTok, those referrals don’t work well in my experience. With referrals, you want them from people that, in this order:

1. You’ve worked directly with in the past

2. A connection to someone that you’ve worked directly with in the past

I’m using the term “worked” loosely here. It doesn’t have to be people that you’ve “worked” with in a corporate setting.

Some of my all-time great referrals have been from college connections that I’ve “worked” with on club events or in classes. My first job as a product manager was through a referral I “worked” with in a club. My second job as a PM came from an engineering classmate I had.

The reason you want people that you’ve “worked” with referring you is that those people can answer questions about what you’re like to work with that the hiring manager or recruiter will definitely have. The last thing you want a referral to give when they’re asked what you’re like to work with is a shrug. 🤷‍♂️

Okay so we’ve covered why referrals work and the quality of referral being important. Now let’s talk about sure-fire techniques I’ve battle-tested for how to land these referrals, with some resources at the end you can download and use…

How do you get a good referral? A step-by-step guide

Read more

I'm using this newsletter to share at least three things l've come across this week to help you build better product. Subscribe so you don't miss these when they come out:

Subscribe now

On to this week's thoughts and updates (note: this week’s update will be brief, mostly because of item #2…):

1. We’re back to text, hopefully that’s good?!

Honestly, I love video as a medium. I love how sharp the video is that you can get these days, the new editing tools, making content super interactive, it’s all great fun. It’s part of the reason I loved my TikTok account and doing videos like this one so much:

@millennialpmThese interviews are way harder than the job (already difficult to do well… 😅) #productmanager #productmanagerlife #techtok #productmanagers #techjobs #jobsearch

Tiktok failed to load.

Enable 3rd party cookies or use another browser

I’m a bit of a perfectionist though (everyone say it at once: NO WAY WE COULDN’T TELL!!!) I want the content to be good, and it just takes a ton of energy for video content to be at the great standard I love. If I didn’t have a full time job and a family, I’d love to make more video and podcast content. But here we are, and frankly it’s easier to write posts on a Thursday night disheveled after FINALLY putting the kids to bed.

So we’re back to writing. Hopefully it’s still great! But let me know if you’d like to see something else.

Speaking of being overwhelmed…

2. Onboarding into a new job is HARD

When I’ve been at a job and think about switching after 15 years of being a product manager, I will admit that I pretty much get this feeling of “UGHGHGHGH that sounds so hard I don’t wanna.”

My subconscious must be with it more than my conscious brain because that feeling is spot on. Onboarding into a new job is super hard and my brain feels like melted ice cream.

Onboarding feels harder at the more senior levels I think because expectations of you are high. Your reputation proceeds you often because you probably got the job via a good referral. And you’re getting older with more life responsibility, so it’s just harder to get immersed and ramp up.

Doing this successfully really comes down to having patience with yourself and your biology. Things will eventually click into place like they did with me today. I kept hearing about a score card and it finally clicked today that the score card is measuring missing data I care deeply about. Your brain will catch up eventually!

So if you’re out there going through this right now, I’m right there with you and I feel for you. It’s hard. There’s no getting around it, the only way is through the pain.

On to more think-y stuff…

3. AI is going to consume lots of energy

I came across this one in my Instagram feed this week:

So let’s be real for a second. The energy production problem has been a problem for some time. My first job out of college was at a power company in Texas. I learned a lot about the people at ERCOT that were scrutinized (rightfully) for the winter storm failures in 2021 and beyond.

When I worked there in 2007, the company was still into burning coal for power because coal was cheap. Natural gas was too expensive, wind was variable and would go out frequently, causing power production levels to drop dangerously close to brownout levels before something else would come online and pick up the slack. Even with nuclear units, it never seemed like we had enough power to satisfy demand.

And this was BEFORE everyone moving to Texas, Tesla, AI, crypto, natural gas prices dropping like a rock, the rise of solar and windmills killing birds. To me, it’s a small miracle everything still works pretty well.

AI though could really push this thing over the edge though. And we have to be really serious about fixing things that have been broken for a long time. I wonder sometimes if our state and city governments are taking seriously, or even have the tools to address this. We need clean, cheap power and we need it fast.

4. Let’s talk about the emotional dial

just has the most 🔥 performance content, I posted this article in a note the other day because of so much professional experience I’ve had with my emotions:

Ultra Successful

Master the Dial: How to Express Emotions as a Leader

Want to listen to this week’s edition of Ultra Successful? Click here or scroll to the bottom of the post to tune in…

Read more

a year ago · 32 likes · 2 comments · Dr. Julie Gurner

I can personally attest to the fact that expressing emotions can definitely be a tool.

For example, I find it really difficult to be a product manager if I’m not into the products I’m building. If I don’t care, I can’t be at my best. So I’ve discovered that making sure I work on things that provide any positive emotional response for me at all is a big step in the right direction.

But when I’m in the job, I find this skill of expressing emotions even more valuable. For example, being vulnerable about my frustrations with leadership expectations I’m not aligned with helps me build trust with my team. They know I share their pain, even if the next thing I’m doing is pushing all of us forward to do the work we’re not excited about.

And sometimes as a product manager, dialing up the anxiety, excitement or anger depending on your team can really move the ball forward on critical work. I find excitement is the best motivator for me personally, but that doesn’t work for everyone. You have to find the mode that works for you and your teams.

That's it for this week!

If you've found this helpful, please consider showing your support by subscribing:

Subscribe now

I’ll be back around this time next week with more useful product manager things!

I'm using this newsletter to share at least three things l've come across this week to help you build better product. Subscribe so you don't miss these when they come out:

Subscribe now

On to this week's thoughts and updates:

1. Everyone Wants a Magic Data Insight Button 🧙‍♂️📊

This week, I ran a poll on LinkedIn asking product managers what tool they most needed — 72% said they wanted AI-powered insights from their data. So I did what any curious PM would do: I built a prototype in Lovable called Trend Whisperer.

You can upload a CSV and get insights in seconds — it's not production-ready, but it's real code, real feedback, and a real reminder that prototyping is moving faster than ever. From idea to prototype in days? That’s the new normal.

2. The Rise of the Super IC and the Future of Work

I keep mentioning Elena Verna because, well, she keeps being right. This week, her post on the shift in full-time employment echoed something I’ve been living myself: the rise of the Super IC.

I've recently gone back to being an individual contributor, and it’s a whole different game now. With AI and better tooling, ICs can operate at massive scale. Product, design, engineering, even go-to-market work — the lines are blurring. The “solo entrepreneur” archetype is creeping into full-time jobs. And guess what? It's working.

3. AI Products: The Basics Still Apply 🤖

Despite all the buzzwords, building great AI products still means building great products. Full stop. Understand your customers. Solve real problems. The same product fundamentals still apply, whether you're working with LLMs, mobile apps, or something else entirely.

No amount of AI magic is going to fix a product no one wants. But pair a sharp understanding of the basics with new tech? Now you’re cooking.

4. The Luxury of Being Offline

Kareem Rahma shared a post on Substack Notes about how attractive it is when people are truly offline — and it hit home. I've been leaning into that lately. Less doomscrolling, more walks. Less noise, more clarity.

We weren’t built to consume an endless firehose of content. And the deeper I go into product and tech, the more I value the quiet time where actual thinking happens. Try it. Your brain might thank you.

Tools of the Week

Google Calendar + Todoist — I just discovered the calendar integration, and I’m obsessed. It lets you assign tasks to blocks of time like a boss.

ChatGPT Canvas + Notes — With memory features improving, I’m now using ChatGPT as my primary thought partner and journal. Way more fluid than traditional note apps.

What I'm Reading / Watching

AI Powered Search — On page 90-ish. Just hit the section on reflective intelligence. A great read if you want to understand AI through the lens of how people search and how products learn.

Children of Dune — Getting back into this sci-fi classic. It’s dense, but if you’ve read the first two, it’s a rewarding continuation.

Formula 1: Drive to Survive — Yes, I’m late to the party. Yes, it’s still amazing. It’s like Succession for racing fans. Highly recommend.

That's it for this week! Thanks everyone for tuning in.

If you've found this helpful, please consider showing your support by subscribing:

Subscribe now

You can now follow along on Spotify or YouTube as well. I’ll be back around this time next week with more useful product manager things!

I'm using this newsletter to share at least three things l've come across this week to help you build better product. Subscribe so you don't miss these when they come out:

Subscribe now

On to this week's thoughts and updates:

At the top: I've created a podcast and video version of these newsletters! I hope you take advantage of these experiences. I hope they're helpful. Hopefully you can listen to this stuff if it's useful on your run, on your commute. Or watching in your side panel while you're doing PRDs.

1. Do things that make your palms sweaty

I've been going through orientation this week and meeting a lot of my new coworkers, and one of the things that came up was an icebreaker: "What was some of the best career advice you've ever gotten?"

I had two. One we can talk about another time. But the other one was that you should do things in your career that make your palms a little sweaty.

You should do things in your career that make you uncomfortable. And I have to be honest, some of what I've been doing lately is a little bit uncomfortable. It's making my palms a little sweaty — working in AI and working in a new space.

I feel like I am living out that advice right now. And it's good advice. You should be thinking about how you can move forward in your career and make yourself feel a little uncomfortable — that means you're learning and growing.

Now, some might interpret that advice as: make sure you're doing things that you don't want to do, or that you have to go into management, or do something specific. You don't.

I think it's really more about personal growth and doing things that you find challenging and interesting and that really light your fire. I feel like that's the journey that I'm on right now and I'm really excited about it!

2. Apple should just give up on Siri

Unpopular opinion alert! But I'm really starting to believe that Apple should just capitulate in the voice assistant race — and capitulate to something like ChatGPT.

You could consider me your tech grandpa because I've been doing this since 2007. 👴🏼 One of the advantages of doing that is that you have a long lens into history. Like, you can remember a lot of stuff that sometimes gets lost in the sands of time.

A lot of people forget that Apple bought Siri in 2010. I had to look this up to get the info right…they bought Siri in 2010 — it wasn't even a homegrown product. And they've been trying to make it great since then, FOREVER ago.

This latest iteration of Siri that's supposed to be coming out is supposed to be incredible. It's supposed to have all of these AI features. And… it hasn't happened.

At some point you have to accept that Apple is, at best, a little behind on the AI side of things. At worst, just not able to deliver the kind of AI experience that users expect now, especially with the advent of ChatGPT, Gemini, DeepSeek, Perplexity… all these new products that are frankly amazing.

You have to lean into the things you're good at. And potentially outsource or yield ground in places you're not. This was all spurred by me doing some research recently on iOS and ChatGPT integration.

One thing I learned is that ChatGPT’s integration in iOS goes beyond just being a release valve for Siri. Previously I thought you’d initiate Siri, and if Siri can’t do something, it punts to ChatGPT.

But if you use the hotword "ChatGPT" before your voice command, you can go straight into ChatGPT. I’ve made advanced voice mode a hot button on my home screen. And it is just so much better than Siri. I don’t know if Siri will ever be that good.

Marques Brownlee did a video on this too, talking about something similar.

Apple has been working on this for 15 years. That’s a long time to be committed to something and not have it work. I don’t know anything about what’s going on inside of Apple, of course, but if I was there, I’d ask: is this ever going to happen or not?

Not to stoke the fires here. I’m just a guy. I work in tech. But ChatGPT as an acquisition target for Apple would make a lot of sense.

Google’s got Gemini. OpenAI’s tech is valuable — not just the model, but the experience. Maybe Apple should go after ChatGPT and OpenAI. Maybe that’s the end game…

3. Why OpenAI is probably raising money on ChatGPT the app — not the models

It’s making more and more sense to me why OpenAI would raise money based on ChatGPT and not the models themselves.

For years there’s been this assumption that whoever figures out AI at scale is sitting on a gold mine. And I think that’s true — but not in the way we expected.

We thought the LLMs would be the gold mine. That they’d generate billions or trillions in value. But turns out… they’re enabling tech. And that value is eroding because the cost of running LLMs is dropping quickly.

Now there are great free models — DeepSeek, Gemini, Meta's new Llama 4 models.

So what is actually valuable then? The ChatGPT experience.

You can see OpenAI separating OpenAI the company from ChatGPT the product. It makes sense. Because the value is not just the model — it’s the app.

And this ties into what we’ve been reading in AI Powered Search as part of the book club (around page 70). Good search experiences parse the query. They don’t just take it at face value. And the processing layer? That’s not happening in the LLM itself. It’s happening in ChatGPT.

Same thing with Perplexity — their secret sauce is in that query processing and ranking layer. ChatGPT is in the same space.

But it’s more than that. ChatGPT is becoming my life assistant. I use it as a daily driver.

Canvas is a great step for documents and notes, but there’s more potential. Tasks doesn’t really work the way I want it to. Projects doesn’t support the full feature set. You can’t use Tasks or Deep Research in Projects right now.

I’m sure they’re working on it. But it just shows you: ChatGPT needs more love. The value is there. And it makes sense they’d raise money on ChatGPT — not the models.

Because model costs are dropping to near zero. But ChatGPT’s value is climbing.

Remember when image generation melted their servers? That’s a sign of real demand. And a reminder that good product takes effort — to help users understand how new tech fits into their lives.

If they just announced "we have image generation," people would’ve been like meh. But they let it go viral — anime, art types, model training — it hit.

That’s what makes ChatGPT so valuable.

4. iOS is getting really complex

This might just be because I’m getting older. (Tech grandpa strikes again.)

But I found myself watching an explainer video just to understand ChatGPT’s new features in iOS. And it really made me realize: iOS is getting complicated.

There are so many features now. Apple does a good job pruning what doesn’t work, but it’s time for a cleanup. Hopefully that’s coming in iOS 19.

5. Heart health update

If you’ve been reading, you know I’ve been on this health journey during my break. Trying to reduce stress and put good habits in place before heading back to work.

I posted a thread about stress on LinkedIn (thanks for all the replies!) — and Tom Hale, CEO of Oura and my former boss, left a comment that stuck with me.

He reminded me that stress is a canary in the coal mine for heart health. VO2 max and cardiovascular age matter just as much.

Mine is about five years older than where it should be…

And I was like DAMN IT!!! I keep trying to fix this. Last time I made progress I was running regularly — six weeks of solid running — and it worked.

But I injured my back. (Tech grandpa strikes again! 👴🏼) So I went back to cycling and weightlifting, and it didn’t help. My cardio age climbed, and stress took over.

I was also doing saunas, but not consistently. Now I’m trying a new routine:

• Zone 2 cycling on off-days

• Weight training once a week

• Daily walks with the dog

• Sauna almost every night

It’s starting to work again!!!

My cardio age is dropping, largely due to a change in my VO2 Max.

Still a ways to go, but it’s trending right. More to come…

That's it for this week! Thanks everyone for tuning in.

If you've found this helpful, please consider showing your support by subscribing:

Subscribe now

You can now follow along on Spotify or YouTube as well. I’ll be back around this time next week with more useful product manager things!

Happy Friday everyone! Welcome back to Ship Happens, your weekly product manager newsletter.

I'm using this newsletter to share at least three things l've come across this week to help you build better product. Subscribe so you don't miss these when they come out:

Subscribe now

On to this week's thoughts and updates:

1. Heading back to work on Monday, I think I have a problem with relaxation…

I don’t know about y’all, but I think I have a relaxation problem. 😅 I’ve been not working for a few weeks now. I definitely feel more recharged and ready to go, but I can’t honestly say I 100% relaxed.

This morning, to help myself feel better (what is wrong with me?!) I made a list of all of the stuff I took care of during my time off. I’ll spare you the full list, but it had 22 ITEMS in it. 🤦🏻‍♂️ And somehow my brain was just like “wow you couldn’t have taken care of the remaining 20 things you wanted to do with your time off? you totally failed man.”

Even today, it’s KILLING ME that this newsletter didn’t go out at 6am this morning. Even though there’s no pressure! And hopefully you enjoy this material enough that it doesn’t really matter when it goes out. I still can’t relax about it.

I’m sure on some level this makes me good at what I do. But it also seems like it would be helpful for me to be able to just lay down and take a nap or something. 😴

Also it’s proof that you can’t completely trust the anxiety-generating machine that is your brain sometimes. Also big thanks to everyone that contributed to this stress thread on Linkedin! I’ve got some great ideas from this that I’m excited to try out as my full workload comes back the next few weeks. Looking forward to being back at work on Monday! 🎉

Let’s back away from my anxieties now and talk more about the future of search…

2. AI Powered Search learnings: AI changes the game in search

Our AI Product Book Club is in full swing this week (you can join here if you’re interested) and I am roughly 50 pages into our book AI Powered Search so far:

For this book I’ve been experimenting with the Blank Page method of taking notes from the folks that run The Knowledge Project. So I have actual notes I can refer to this time!

(A small tangent: I’ve been suspecting that just highlighting books and articles isn’t the way for me to learn anything from what I read. It turns out I was right 😅)

So here’s what I’ve taken away from this book to date:

1. LLMs have the capability to make search much more powerful, and users expect that. The advent of products like Perplexity with LLMs at their core have significantly increased user’s expectations from search. And it has made feeding signals, structured or unstructured, back into search to measure effectiveness to users more possible than ever. We’ll talk more about concrete examples in a second, but LLMs like ChatGPT can make previously near-impossible problems to solve more possible to resolve. One specific example they talk about in the book:

2. Unstructured data can now be represented more correctly as hyper-structured data that can be parsed, categorized and searched against with the use of an LLM. Previously, if you had a pile of documents that had no clear structure being followed, this would be considered unstructured data and ignored. An example of this is lecture notes that instructors would upload at Udemy, my previous workplace. Before, we had thought these documents were un-indexable and not usable for search. Especially if they were non-text docs like pictures or video. But now an LLM can parse these documents, structure their data in vector representations, and you can use vector similarity functions in search to search across them with a query. All of a sudden, something that would have taken quarters to do is significantly shrunk in scope with an LLM. Speaking of vector representations:

3. You can use the LLMs to create those vector embeddings at a large scale in a much shorter time frame. This is what I think it’s going to take me some time to process with LLMs, the fact that so much can be done much faster and at scale than was previously possible. Creating things like vector embeddings can be done much faster and at a large scale, making hyper-structured data available to search much sooner than multiple quarters. LLMs can make the vector embeddings much faster and process queries to find similarity much faster and more accurately than before, which makes search way more useful, faster.

4. The more traditional way of thinking about search & recommendations separately is disintegrating. Historically, I’ve seen the Search and Recommendations teams be separate within organizations. But with recommendations increasingly being needed within search, search algos driving recommendations, and LLMs becoming more and more central to them both, it’s making more and more sense that they are one team. Which has actually made sense to me for a while, but the book has crystallized that more.

So the short answer to the question “Will AI change the way search works forever?” is likely yes lol. So much that I personally had thought of as limitations to search being great are no longer limitations which is SO EXCITING.

If you want to learn more, join us in the book club! We’re just scratching the surface of AI Powered Search, and I’m always on the lookout for our next book to check out.

Let’s shift gears and talk about AI-driven development…

3. Using Lovable & Cursor has me wondering about the prevalence of AI-driven development & tools

I kicked off this poll on Linkedin yesterday:

The results are trickling in, but it does seem like development in partnership with an AI copilot is on the rise. There are still some people holding out as you can see with that 30% figure, but the majority seem to be using AI for some professional coding tasks. Which is not surprising given how powerful these tools are.

The leading product for professional use seems to be Cursor, although Lovable just came out with the capability to edit your projects with Lovable this week 👀

So we might be seeing more changes to the development tooling landscape over the next few months.

It also seems like it depends on the nature of work. One of my previous coworkers had this to say about it:

Have our AI overlords taken over professional software development then? Not yet. And the role of the professional engineer is likely not going way, at least anytime soon.

However, much of what we thought needed a professional to do is in fact going away though. The next few years should hold some exciting developments here (see what I did there?! #dadjokes)

You can follow the thread here if you’re curious. The next few things are mostly learning recommendations, so fire up your bookmarking app if you need to!

3. Recommended video: Andrej Karpathy’s YouTube channel

I have to say that Andrej Karpathy’s YouTube channel is excellent if you’re trying to dive deeper into the world of AI and LLM’s. I’m sure many folks have already gone deep here, but I just discovered it and it has become my go-to watch any time I have some free space to learn more about LLMs and AI.

The video I’ve linked to at the top is a great starting point. He’s a very engaging speaker and provides some great examples on how LLMs work, what their pitfalls are, and what to expect in the future. It’s 60 minutes 🤯 but don’t let that scare you! It’s a very engaging watch and flies by.

4. Growth newsletter recommendation: Elena’s Growth Scoop

Elena's Growth Scoop

Subscription churn metrics and benchmarks for operators.

Sponsored by Churnkey - the platform that helps companies save 20–40% of subscription revenue that would otherwise be lost to churn. Used by best-in-class companies like Jasper and Superhuman to proactively improve retention. Implement Churnkey today →…

Read more

a year ago · 60 likes · 6 comments · Elena Verna

If you’re not subscribed to Elena’s Growth Scoop already…what are you doing?! I’ve been a Growth nerd for sometime, even if you’re not in Growth per se there’s a lot of value in learning about it. Many product managers spend so much time focused on delivering core value, and then assume Marketing or someone else will drive growth. That is a mistake!!! True growth comes from creating value AND making sure users access, understand, leverage that value and tell others about it.

The edition above I found super useful about how best to measure churn and how to address it. Highly recommend her work if you’re building products you actually want people to use. 🤣

5. Recommended sci-fi fiction: Infinity Gate + Echo of Worlds

Okay last thing and I’ll let you go…

The Infinity Gate + Echo of Words series is pretty great if you’re into sci-fi, AI and all things tech. I just finished this last night (I read fiction before bed usually), it’s pretty fantastic. Based on the premise that people figure out how to traverse dimensions where there are infinite versions of Earth. Really imaginative, would give it a go!

That's it for this week!

If you've found this helpful, please consider showing your support by subscribing:

Subscribe now

I’ll be back around this time next week with more useful product manager things!

I'm using this newsletter to share at least three things l've come across this week to help you build better product. Subscribe so you don't miss these when they come out:

Subscribe now

On to this week's thoughts and updates:

1. AI Book Club: AI Powered Search!

I’m starting my new gig in AI in a couple weeks! I am super pumped to be working on building the future, but a little anxious about understanding the fundamentals more deeply around AI, search and their respective futures.

My coworker Natalie recommended reading AI Powered Search and so we entered into an unofficial book club between us, committing to finishing 100 pages by the end of March.

I then posted about it on Linkedin and it sounds like other folks wanted to participate too..

So we’re doing a book club! Let me know if you’d like to participate.

I thought about creating a Substack Chat, but it looks like there are lots of tools out there for this that are pretty cumbersome (Linkedin, Slack, iMessage, WhatsApp, Substack…) It got me thinking about having a social book club app which is perfect for vibe coding! The perfect segue into…

2. Vibe coding is a total trip, and probably here to stay

If you haven’t heard about vibe coding yet, I’d recommend pausing and reading this:

KIRUPA 🍊

"Vibe" Coding today, just Coding tomorrow! 🚀

Hi everybody…

Read more

a year ago · 3 likes · KIRUPA 🍊

Vibe coding with an AI tool like Lovable (my tool of choice), is quite the experience. You get a chat window to interface with Lovable, and you basically write out what you want to make in prompts that Lovable executes against. Then Lovable uses AI to write out code for pretty much anything you need.

I’ve had previous experience writing code with ChatGPT as my partner in 2023, and that honestly was a struggle. I had to provide ChatGPT lots of context, and often it would do integrations the wrong way with old knowledge it had been trained on.

Now, likely with the addition of RAG and updates to the models used along with some “magic” behind the scenes, For the most part I can tell Lovable what I need and it just does it. 🤯

I’ve been using it for two different products:

1. A quiz game called Quizzy (get Quizzy with it, get it?!) that my brother-in-law came up with. It creates timed quizzes based on popular TV shows that you can take shortly after an episode has come out.

   

2. A social book club app, name is WIP, where you can set up your own book club by picking a book, inviting friends, setting milestones and then scheduling a debrief over Zoom at the end. Perfect for solving my book club problem above!

I have to say, using AI in this way to build products in a matter of a couple of days is totally amazing. I finally feel like this promise everyone talks about where AI can build digital products for you is actually a reality. We were very far from that in 2023.

But vibe coding definitely comes with some unexpected learnings and caveats:

1. My PM skills are now more important than ever. I found myself really articulating use cases, specifics on requirements, and digging in deep on requirements. I also still got stuck on what I should use these tools to build and why. If anything, these tools seem to accentuate the need to figure out what to build and why, BUT…

2. Being a PM takes on a different shape. Now you can build really cheap and test ideas quickly instead of having to do lots of things upfront before even seeing a line of code. I’m not sure what it all means yet for the way product managers and other roles work going forward, but on initial glance it looks like lots of lines are going to blur and needed skillsets are going to change. For PMs, the need for docs and upfront designs might not be as critical if you can prototype and V1 fast with AI. Actually this article from Julie Zhou is a great meditation on that topic. The future of our role and what will be important is likely to change, as it will change for designers, engineers, data scientists, really all of us.

3. Vibe coding platforms are very much like junior developers. You really have to spend a lot of time checking their work and QAing. So you might not be writing code, but you still have to have a good sense for what the requirements are, where the edge cases are, what’s working or not working and why, and how you could potentially solve it. This also works its way into prompts, where you really have to be specific about what you need and how to craft something in a way where you get what you need.

The future is very bright, and vibe coding or Prompt Driven Development (PDD) as has so aptly highlighted, is likely here to stay and the future of how we build products.

I can’t wait to share with y’all more about Quizzy and the book club product! It’s coming along super quick, might have something working to share next week. 🤞Next let’s talk about what’s melting down OpenAI’s servers…

3. GPT 4o’s image generation leap forward is amazing, concerning, and oddly feels like it should have happened years ago?

If you haven’t spend the last few days messing with ChatGPT’s image generation model, why not?! It has pretty much melted down OpenAI’s servers and taken the world by storm.

I have admittedly spent way too much time making lots of fun images to see what the model can do and it is totally shocking how well it can take images and inputs and generate something striking in terms of completeness relative to prompt with detail.

Here are some examples of what I made, the prompts are in the captions:

Here’s one of my pup Einstein in Pet Vogue:

It’s amazing the level of detail and correct interpretation of the prompts given. It even gets a very good approximation of the text in things like my shirt, although I do have to correct it roughly half of the time. Sure Midjourney is great at a lot of this stuff, but the level of accuracy with image representation for things like depth has taken a leap forward, and it’s just captures + displays intended artwork so well it’s mind blowing. It unlocks a whole world of creation for so many more people on the planet, including myself who thought making this kind of stuff myself would never be possible.

It’s concerning in many ways, the key ones to me are what it does to existing jobs and what it does to creating art that is truly unique and human. On jobs, I wonder about if I will really need a graphic designer, visual illustrator or photo editor ever again. It makes posters, visually treats UI, creates comic books, like what the hell?! We’re probably looking at a world where people in those jobs will have to change their skills and the nature of their work.

On the nature of art, this is hard to suss out the impact of, I’d imaging we’re going to see artists using these new tools to make this faster and more unique like this epic Lord of the Rings video. There will probably be another explosion of creativity similar to the effects of the internet and mobile, but at a scale much more mind-bending than even that. There will probably be some lower grade art, lots of copy cats and rip-offs and things like that. Just like there were fart apps in the App Store. 🤝

But for the hundreds or thousands of fart app equivalent art pieces, there will be singular pieces of art that would never have been possible without AI I think. Only time will tell.

Speaking of fart apps, here’s something you probably knew that actually is drivel and worth getting rid of…

4. Social media is aging our brains quickly, screen time isn’t relaxing?!

I came across this on Instagram the other day that was striking:

A decade of cognitive aging gone after two weeks without the internet?! That sounds like total BS, but it aligns with something I’ve discovered about my own stress that I wrote about on Linkedin yesterday:

I’ve seen this play out time and time again, where when I watch TV or look at my phone a bunch like I did yesterday, no restorative time:

But when I do things like disconnect and do some yoga or meditation, I get restorative time almost guaranteed on my Oura Ring:

I used to laugh when people would say that social media is like smoking, but as we’re learning more about it I’m starting to believe there’s some truth in it.

I’m not a total nut about it, I don’t think I’d advocate for cutting social media out completely. But definitely stepping away from it for good periods of time, or finding social media like Substack or Linkedin that’s more vegetables than candy, seems to be helping in real life. Hopefully it gives you some food for thought in terms of how we can all show up better for work and for life.

That's it for this week!

If you've found this helpful, please consider showing your support by subscribing:

Subscribe now

I’ll be back around this time next week with more useful product manager things!

I'm using this newsletter to share at least three things l've come across this week to help you build better product. Subscribe so you don't miss these when they come out:

Subscribe now

One announcement: there will be no newsletter next week. Taking the kids to Florida 🐊 for spring break. Wish us luck!

On to this week's thoughts and updates:

1. Goodbye Udemy, hello AI Copilot at Procore!

I mentioned last week that an end of an era was nigh for me, and so it has come to pass that I am leaving Udemy on Monday after almost 4 years at the company. (Would have been 4 years officially in April. Missed it by a hair!)

I’m so grateful to have worked with so many bright folks in Austin and the Bay Area on many growth, conversion and experience challenges at Udemy. I got to reunite with many people I loved working with at Vrbo! We IPO’d, my younger son was born during that time also, many adventures were had. Thank you to everyone I got a chance to work with. Udemy will always have a special place in my heart. 💜

I’m super excited to be joining the team at Procore in April! I’ll be working on their AI Copilot experience, leveraging a lot of what I’ve learned here and building my own stuff like this resume reviewer for product managers:

Millennial Product Manager

Free Resume Coach for Product Managers (for now! 👀)

Resumes are critical to the job search. Mistakes on your resume can be the difference between getting that recruiter reach-out or not. You need a good resume to at minimum open that first door with a recruiter…

Read more

a year ago · 3 likes · Ajay Waghray

It’s a really exciting chance to work on something different from growth & conversion, and of course in LLMs and AI, an area I’m super passionate about as y’all can see from my musings here. Off to new adventures!

As such, I’ve been learning and re-learning some important things about the job search over the last 30 days or so. So I thought it’d be good to share those learnings below for you in case you’re facing a job search, either now or down the road.

Below are some thoughts about the process of changing jobs and what I’ve learned about it over the last 15 years that you can apply for yourself.

2. Don’t bother applying online. Get a referral.

Another hiring manager and I were comparing notes about what he was seeing as part of the recruiting process for an open role he had.

He said that when he opened the role, he had over 400 applicants in a little over 24 hours. But in reality, 5-10 of those applicants were actually qualified with good resumes. The rest probably were either Easy Applies or AI bots submitting on behalf of candidates with AI generated resumes.

When you apply online, just remember you’re competing with of course AI-submitted candidates, but also in general nameless, faceless people. Trust between you and the company, recruiter and hiring manager will start low. You’ll have to build it brick by brick as a candidate, which leaves very little room for error.

When you have the all-powerful referral, especially if it’s a really quality one from your network of people you’ve actually worked with or at least know you really well, you have a big leg up over the faceless AI-generated masses.

This makes a strong case for referrals in your job search, but just to add a kicker and bring this home, I asked Perplexity about how best to look for jobs. It told me that 80% of jobs get filled via referrals before they even hit a job board. So, ya know, get a referral!!!

How do you get a great referral? You’re in luck! I made a series of videos about it on TikTok, starting with this one:

@millennialpmI wouldnt even bother applying without a referral in this job market, make sure you get one 🤝 next video we’ll talk more about how to get a great one #productmanager #techtok #techjobs #jobsearch #corporatelife #productmanagerinterview

Tiktok failed to load.

Enable 3rd party cookies or use another browser

3. Build your brand

Okay, this sounds like one of those super-gross things that business influencers say. But go with me here for a minute.

Remember how we were talking about building trust above? Well even if you have a referral, it’s hard to get into people’s heads during the interview process, even if you’re a great interviewer. I’ve been told I’m pretty good at interviewing and reading people, but even for me interviewing can be a bit more art than science.

So what’s another way to get into someone’s head aside from the interview process? One great way is from the referral you undoubtedly got from above RIGHT?!

Another great way though is to read, watch or listen to anything the candidate has put out into the world. This includes things like this Substack here! And OF COURSE any actual product you’ve built.

So get yourself and your brand out there. Build your TikTok following, write on Substack, make a podcast (I’ve done all of those things!) Be on someone else’s podcast, post on Linkedin more…get your thoughts out there! I can tell you for sure that the last three jobs I’ve gotten have been influenced by my thoughts out on the internet. Get out and make stuff! Below is a more advanced variation on that…

4. Side hustle on something you want to become your main hustle

Here’s an excellent video from Prof G on side hustles:

Prof G can be hit or miss I think. But on the point of side hustles being a temporary means to make something else your main hustle, I think that’s on point.

All of my side hustles in the last few years: the AI resume reviewer, building an AI-generated trips app, writing this newsletter…they’re all off the beaten path of what I did before (mostly growth & conversion.) But they all helped me transition to the new career path I’m on working in engagement and AI.

So if you have the time and bandwidth, I’d highly recommend doing a side hustle, mostly for learning and getting versed in the things you want to do ultimately in your main hustle. It helps you stay sharp and gets you to where you want to be.

There are so many new tools now to make something great too! Cursor I love and now use as my main IDE. But things like Lovable, Replit etc. all sound amazing. I’ll be testing them out during my time off!

5. Practice interviewing with ChatGPT Advanced Voice Mode

OMG I love ChatGPT Advance Voice Mode SO MUCH. I spent HOURS with it prepping for interviews. Here is what I did to get the most use out of it:

1. Upload your resume for context so it knows your background and experience.

2. Give it a link to the role you’re interviewing for.

3. Give it links to Linkedin profiles for the people you’re interviewing with.

4. Paste in any notes that you have from your referral, the recruiter, and any interviews you’ve had, along with notes about the company, their website, etc.

5. Use a prompt like this one before you start advanced voice mode:

Let’s role play for the role I’ve included above. What kinds of questions would I be asked? Please provide feedback on my answers as well. I’ll trigger voice mode to mock interview.

I’d say I spent 5-10 hours mock interviewing this way and it made me feel so much more comfortable walking into interviews. Do it, thank me later.

6. Negotiate like a boss

If you haven’t read this book yet and don’t have immigrant parents like I did to teach you how to negotiate like a total boss, add Never Split the Difference to your Amazon cart right now:

I read this before I started my 2025 job search journey and was head-nodding through almost all of it. It reminded me so much of watching my dad negotiate buying a car…or pretty much anything he bought. Everything was negotiable, he was always looking for a deal and was polite about it.

Much of what I learned from my dad and what you’ll learn in this book will be critical in a job search. Negotiation of compensation with a job starts from the minute you apply, especially now when salary ranges are posted on jobs. And the negotiation continues right up until you sign on the dotted line. How you negotiate can be the difference between a life-changing job change and a big disappointment. So make the investment to get better at it with this book.

Switching gears on the final leg here…

7. I think I’m finally starting to crack my stress equation

We are starting to crack the code on stress! Here’s what I’m learning might actually work for me according to my Daytime Stress chart from Oura Ring:

1. I might need connection to feel relaxed and at my best. The chart above was from this past Sunday, where I spent much of the day with family and my kids. Connection is the key to restorative time for me.

2. Meditation really resets me into a more restored mode. I’ve been hitting the Calm app a ton the last few days and it is delivering at minimum more relaxed moments, if not some restorative time. Brings the heart rate down and keeps me connected to the present moment.

3. I’m not convinced that watching TV or scrolling TikTok is that relaxing for me. I don’t seem to get into a restorative state with those activities. More experimentation is needed but I think that jives with earlier posts.

4. I might try more reading and walking. I wonder if the screen time and lack of movement is building up stress. More on these in the next edition.

So if you’re on a journey to live longer, happier and healthier and are thinking about managing stress to do it, here are some pointers that help.

That's it for this week!

If you've found this helpful, please consider showing your support by subscribing:

Subscribe now

And please consider supporting by becoming a paid subscriber at $5/month or $50/year. I don’t require it to get great value out of the newsletter, but it does provide great motivation to keep going. So consider going paid, you can participate in the paid subscriber chat and keep this sucker going.

I’ll be back around this time next week with more useful product manager things!

Happy Friday everyone! Welcome back to Ship Happens, your weekly product manager newsletter. Unfortunately, I’ll be skipping the newsletter this week. We’ll be picking back up next week.

It is indeed the end of an era for me. 😭 I’m excited about what’s next, which I’ll be sharing more about next week. 🚀 But for now, I’m going to try to get some sleep lol.

Looking forward to more fun updates next week!

I'm using this newsletter to share at least three things l've come across this week to help you build better product. Subscribe so you don't miss these when they come out:

Subscribe now

On to this week's thoughts and updates!

1. Claude Code looks insane 👀

I just have to reiterate here that Claude Code looks really freaking awesome. Here’s a video of it if you didn’t catch their announcement:

I have to say that a really smart design choice here was to go with using the terminal as the primary entry point and platform for Claude Code. With the command line, you can have more direct access to the code base and other functions needed to address issues.

More and more tools keep popping up to either make things like development or design faster for seasoned pros or make development/design easier to access to more people. I for one am loving how quickly the way we build things is changing for the better.

It does seem like a future world where we are all able to scale 10x of our capacity to do great work is heading our way soon.

Subscribe now

2. Amazon’s Alexa Plus Play, The Power of Complementary Investments

Alexa got a big upgrade this week that I have been super excited about:

It seems confirmed that Anthropic’s Claude, ChatGPT’s competitor funded by Amazon, is powering the new Alexa experience.

Amazon’s pumped $8 billion into Anthropic, which seems to be pennies compared to what’s going to be spent on AI by many players, OpenAI included.

Clearly Amazon’s bigger play with this investment was to not only compete with OpenAI, but also use Anthropic to build the foundation for it’s own AI ambitions down the road.

Just goes to show once again that there is more than one way to build new tech. You can acquire it (that could still happen with Anthropic by Amazon), you could fund it and partner with it, or you can build it in house but you need the expertise. So, often companies buy, partner or invest to make something like this happen, faster and more efficiently.

Another genius move by Amazon.

3. Comet vs. Operator: Why Performance, Design Decisions Matter

MOAR NEWS!! Perplexity is launching Comet, an agentic web browser designed for Perplexity to automate tasks on behalf of its users:

Earlier this year I tried Operator from OpenAI and I have to say it was the slowest thing on earth to use. It was a hosted browser on the cloud that would be used to complete tasks for users.

I’m sure it’s gotten better and will get better, but I also started to question why you would want to host a browser to run agent tasks instead of run right on someone’s hardware.

Well, Perplexity took care of that with Comet 🚀 By creating their own browser, now tasks can run local on the machine and the user can interact directly to make any changes as needed.

So when you’re thinking about making your own apps or experiences, think hard about how you design the experience and technology stack. If you can land on a solution that’s better because you thought deeply about the design problem, it can be the difference between being a punch-line in someone’s write-up and cornering the AI browser agent market.

Also, a good rule of thumb: having your software be more performant is almost always a good thing. Slow = bad.

4. My stress numbers got all torqued from being sick

Last week I talked about going on a journey to reduce my stress levels as a way to improve my sleep and recovery.

Well, all of that got torpedoed by having a cold! 🤧 So we’ll have to try again now that I’m on the mend.

Meanwhile, it was interesting to see on my Oura Ring stress chart how much different things affected my stress level while trying to get better.

For example, I had no idea how much steroids will jack up your stress levels. The day I took them, I was in a stressed zone according to my Oura Ring for 9 hours?!

I was also surprised to see my first full night’s sleep in a while result in a lower readiness score. Probably just a lot of built up stress and sleep debt. We’ll see how the next few nights shake out.

But make sure when you get sick you get lots of rest!!

That's it for this week!

If you've found this helpful, please consider showing your support by subscribing:

Subscribe now

And please consider supporting by becoming a paid subscriber at $5/month or $50/year. I don’t require it to get great value out of the newsletter, but it does provide great motivation to keep going. So consider going paid, you can participate in the paid subscriber chat and keep this sucker going.

I’ll be back around this time next week with more useful product manager things!