Ga offline met de app Player FM !
The Future of Onboard AI and The World’s Best Water Bottle
Manage episode 423513653 series 2733
In this Marketing Over Coffee:
Learn about what came out of WWDC, how to market better on LinkedIn and more!
Brought to you by our sponsors: Wix Studio and NetSuite
Apple WWDC
Entering the age of on device AI
7:04 Wix Studio is the web platform that gives agencies and enterprises the end-to-end efficiency to design, develop and deliver exactly the way they want to!
iPhone mirroring
Apple has such great UX, why is Apple Music on the Mac not there? But they are paying artists…
Transitioning from SK Ad Kit to Ad Attribution Kit
iOS 18 to add RCS – Rich Communications Text Services
14:28 NetSuite is the number one cloud financial system, bringing accounting, financial management, inventory, HR, into ONE platform, and ONE source of truth.
Updated Unofficial Guide to LinkedIn Algorithm
Gear Watch – Owala, the new best in class Water Bottle, WorkOutDoors for Running
SparkTogether coming up in Seattle
Gen AI Course Updates done: Special Discount on the newest Generative AI for Marketing Course! Hands on excercises to put AI to work for you! USE CODE MOC now!
Join John, Chris and Katie on threads, or on LinkedIn: Chris, John, and Katie
Sign up for the Marketing Over Coffee Newsletter to get early access!
Our theme song is Mellow G by Fonkmasters.
Machine-Generated Transcript
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.
Speaker 1 – 00:07
This is Marketing over Coffee with Christopher Penn and John Wall.
John Wall – 00:14
Good morning. Welcome to Marketing over Coffee. I’m John Wall.
Christopher Penn – 00:17
I’m Christopher Penn.
John Wall – 00:19
And we have all kinds of stuff this week. It’s been crazy, there’s a bunch of things happening. The biggest one we’ll lead off with is Apple’s WWDC, their Worldwide Developers Conference. They roll out some new stuff, but then there’s a bunch of stuff on AI and all kinds of stuff going. So what’s — what was the biggest thing for you? What were you like, “Oh my God, this changes everything?”
Christopher Penn – 00:37
So it’s interesting. They had the big keynote, right? And everyone’s like, “Oh awesome, flashy stuff” like that. But that’s not where the details are. The details are actually in the session because it’s a dev conference, it’s a developer conference. It’s not for the general public. I mean, a lot of people tune into the big keynote to see the stuff, but all the — the really cool/very subtle things are in the individual sessions. The big one to me — and this is very nerdy — the big one to me is the way Apple has architected its AI, which they have cleverly branded Apple intelligence in a couple of different ways. Most of it, from the system architecture diagrams, looks like it runs on the device itself.
So we’ve been seeing, really since like iPhone 13, the number of neural cores in the neural engine on iPhones and Macs and stuff has been getting bigger and bigger. In fact, the new M4 chips that just came out for the new iPads, the previous M3 had the — the neural engine was able to do 18 trillion operations per second. The next generation, the M4 chip, the baseline is 38 trillion operations per second. So they have effectively doubled — more than doubled — the amount of processing these things can do that allows Apple’s custom models to run entirely on the device. So all the cool stuff they’re showing — summarize this document, make a silly image and stuff — all that’s going to run locally, which means that your data is not leaving your device, it’s not going to some unknown cloud provider.
Christopher Penn – 02:10
They announced the integration with ChatGPT, which I thought was very interesting because the way they implemented it is super clunky on purpose. Every time you want to use ChatGPT, you have to confirm like, “Yeah, you’re about to send data to a third party, do you really want to do that?” Which I think is so smart because, A, it allows them to easily swap out another provider — and they’ve said Google Gemini support will be coming at some point — and B, it reminds people, “Yeah, when you send your data to OpenAI or whatever, it’s in most contexts, it’s not private.” So I thought that was really cool.
What was interesting in the Core ML session is they were showing you directions on how you can load your own — how you can convert your models from floating point 16 tensors to Core ML, which is their architecture, and they showed how you prepare it and all this stuff. And then said, as part of your app, you can load this model into the phone or the device and run it locally on device. When you think about it, as long as someone’s really good at model making, that’s huge, because such a big part of the cost of AI these days for a lot of these companies is you’ve got to pay the OpenAI tax or the Google tax, or whatever, to send your stuff to their hardware to process it and come back.
Christopher Penn – 03:33
If your model could fit inside of Core ML, if you can be really efficient and smart about it and make a very dedicated-purpose model — you pay nothing in technology fees because you’re just using the existing Apple hardware.
John Wall – 03:46
Yeah. And we’ve talked about this, even like you said, back when 13 came out, we had said that the idea with this is that it could run these models locally. But, yeah, it’s just kind of crazy now that it’s here. I mean, so is this — the idea is you have — it’s nodes. I mean, you have machine learning running at every single point out in the field, distributed like you said, it doesn’t have to phone home at all. And just a side note too, I did — Elon Musk, went crazy and said that if ChatGPT is baked in, he’s not going to allow Apple devices in the factory or anywhere.
John Wall – 04:19
And the article made clear the point that — well, obviously he missed the idea, which is that it has that button that screams at you saying, “You’re sending this data to a third party,” which, yeah, that is really interesting too, kind of sowing seeds of discontent there. You’re basically telling the world that — think about what you do before you push that button because it’s leaving the device.
Christopher Penn – 04:38
Yep. Generally speaking, Elon Musk is such a petty sore loser that anything that’s a good idea that he craps on is probably going to do really well. So like, it’s like the — I guess, the opposite indicator. If he loves something, like, yeah, that’s garbage.
John Wall – 04:56
You’ve got to stay away from it.
Christopher Penn – 04:57
Yeah, I mean, to be fair, when you look at the design of the Cybertrucks, and you’re like — and they look at like an Apple device, like, okay, one of these things is not like the other. And I thought this was really interesting about the market opportunity for companies with on-device AI. Again, if you are good at building or — you know someone who’s good at building a small, very finely tuned single-purpose model, and you can cram that into — you know, maybe a one billion parameter model, maybe a 150 million parameter model, maybe you have a model that is just trained on your podcast and nothing else. It can’t even tell you what color the sky is, but boy does it know your podcast.
Christopher Penn – 05:40
You would integrate that model into your app in the App Store and have conversational AI in your app that is specifically focused, that gives someone the generative AI experience. I can talk to this podcast app and get, “What is Marketing over Coffee done with mobile marketing?” And because it knows the index, and you have the data and stuff like that, it will do a great job of that. If you ask, “What do you know about the In-Ear Insights podcast?” “I have no idea. I’ve never heard of that podcast and I’m not going to go look because that’s not what I’m trained on.”
This gives people the ability to create rich AI experiences that are baked into the app, baked into hard — run on the hardware, need no Internet other than the data that’s there, and could do really well. So if you have at a shopping guide, and you’re like, “Maybe this guide really knows women’s blouses and nothing else.” You could have a really rich shopping experience using generative AI that’s so tightly focused and never needs the Internet. And so I think there’s a reasonable possibility that an app developer skilled with making AI models will be able to create a huge billion-dollar company app for the iPhone ecosystem that has a server AI cost of $0.
John Wall – 07:00
Alright, yeah, there’s a bunch of other stuff too. We’ve got to peel out of this too, but before we do, we just want to take a second. We want to thank Wix Studio for their support of Marketing over Coffee. I’ve only got one minute to tell you about Wix Studio, the web platform for agencies and enterprises. Whether you manage 10 sites or 1,000, here are a few things you can do from start to finish in a minute or less on studio: set up native marketing integrations in a click; reuse templates, widgets, and sections across sites;
John Wall – 07:26
create a seamless handover by adding tutorials, guides, and more to client dashboards; work on the same canvas at the same time with all your team members; and leverage best-in-class SEO defaults like server-side rendering and automated structured data markup across all your Wix sites. Time’s up, but the list keeps going. Step into Wix Studio and see for yourself. Check it out over at wixstudio.com and we thank them for their support of the show. iPhone mirroring was one thing that I saw, that’s something that — on the Android side is not a concern. But, yeah, that was very cool as far as — especially, you made a point about marketers that have phone apps now being able to do it on the desktop.
John Wall – 08:03
I mean, that is game-changing as far as making headaches go away and being able to type and cut and paste and all that kind of stuff.
Christopher Penn – 08:09
Oh, yeah. I was blown away by that, where you have the virtual iPhone on the desktop, and you drag a file from — from your desktop into the social media management app on the iPhone, as it was — as though it was a local file. Stuff like, okay, that is — because phone mirroring has existed for a while. You’ve had the ability to mirror phones, but it’s always just like, “Okay, there’s the screen of your phone on the desktop, but now you’ve got to pick up your phone and use it.” This is making it a true virtual environment where you can control every aspect of that device from the desktop, and it does not have, like, the crazy lag of, VNC or — you know, the many other things. Such a cool —
Christopher Penn – 08:49
Here’s the thing about Apple, and I’ve been saying this a lot lately, they are not first to anything. They did not have the first personal computer. They did not have the first smartwatch. They don’t have the first smartphone. They did not have the first anything. They have the best. They spend so much time obsessing on the customer experience and making sure that it is flawless and polished that even though there are other market entrants, they simply do a better job having a smooth experience so that someone like our parents — there’s no friction. Like, you don’t have to explain to a parent or a young child, like, “Here’s how you do this, and you’ve got to do the step and open this DLL file and all this.” No, it’s like, “Okay, just follow the on-screen instructions and it will do it.”
Christopher Penn – 09:33
I mean, you’ve — you’ve had that experience with relatives.
John Wall – 09:37
Oh my God, yeah. I have transitioned everybody to Apple just because — to make it all go away. It’s a walled garden that can — can hold them well. But so this does — I had this further down in the notes — but this brings up a sore point of why is Apple Music this redheaded stepchild of this whole thing? The desktop version of Apple Music is just missing so many features and just doesn’t — has so many things going wrong. It’s weird — it’s a cord enough — but I had — over on Threads, we were kicking this around a lot, and there was a wise observation that somebody made that, like, look, Apple Music is just there to sell more phones. Like, it doesn’t compete with Spotify or Tidal or any of those things.
John Wall – 10:13
It’s just that they know that’s one more piece of lock-in to sell more phones. And when you’re selling billions of dollars of phones, then that’s fine. So yeah, I’ll throw that out to the crowd. If anybody else has Spotify versus Apple Music opinions, I’m sure everybody does, we’ll hear that. But yeah, I took it for a test run for a week, and after a week I was like, “Well, I still like the Spotify interface better.”
Christopher Penn – 10:35
Marcus Brownlee did an interview with Tim Cook, which is, like, the first time, I think, a lot of the social media influencers have been doing CEO interviews. I saw Justine Ezarik did one as well. And Tim Cook says something very interesting, which is obvious and true, but you don’t think about it in today’s environment. He said Apple’s business model is not contingent on making you spend more time on your device. Apple’s business model is to sell these devices, and — and yes, the services that go along with them, but it is not to make you an addict to your device the way that Instagram, like, literally has to make you an addict for their business model to work. The more time you spend scrolling, the more time you spend consuming ads, the more money they make. Apple doesn’t do that.
Christopher Penn – 11:12
And so for those services, yeah, it’s not — you don’t need to spend as much time, but also, Spotify is kind of a one-trick pony. It does one thing, which is play music and audio, whereas Apple does many things. So yeah, every individual app probably is not going to be as perfected as a company whose sole focus is that thing. It took Apple, what, five years to get a decent map app on their phone? Right now, Apple Maps is good, it’s very good, but it — you know, they ceded a lot of territory initially to Google Maps because the original Apple Maps sucked.
John Wall – 11:45
Yeah. And then if you want to go full audio geek, Apple has a bunch of stuff in Atmos, so if you’re — you know, top-end audio stuff, you can get that. But that opens up a whole ‘nother audio file door of, just because it’s a new format, does it sound better, yes or no? And it really all comes down to individual listening, like, you have to just go listen to all these tracks and figure out which one is better. Unfortunately, there’s no easy solution as far — and the equalizer is the other one that goes in there. Spotify does have an equalizer even on the phone app, which iPhone doesn’t have yet, but they do have on the desktop.
Christopher Penn – 12:19
Yeah. The other big thing for marketers is they’re moving on from SKAdNetwork to their new ad attribution kit. So this is new software within the iPhone, the iOS ecosystem, to help advertisers understand better how the ads are working and things. But it’s very interesting, all the data is aggregated, so you get no user-level data whatsoever in any way. So this makes it compliant, obviously, with all of the major regulatory requirements for advertising. But it’s also interesting the way they positioned it in the tech session on it. You get less data, the smaller the audience is that you’re advertising to. You get more data as your audience gets bigger because there’s less chance of being able to re-identify someone from the ad data you get. So they were showing off, as your audience gets to really big sizes, yeah.
Christopher Penn – 13:08
Now you can have more things like gender and age and stuff like that, whereas you have a very small audience, like, “Nope, you had people, but you had people download your app.”
John Wall – 13:16
Yeah, they blocked that off. That makes sense because that is so easy to game. You can just start to individually tag people, that makes sense. Oh, yeah. Another interesting thing on the tracking side too, is with iOS 18, they’re going to be supporting RCS — this Rich Communication Services — which Android has had forever and Google has put forth as a text message set of standards. And so the idea is now we won’t have this weird gap between you’re sending GIFs and emoji responses that are tied to the bubble and all that kind of stuff via Apple that you couldn’t do on the other side. And they’re saying that there’s actually more tracking data on that too, you’ll actually get analytics as far as what’s happening with text.
John Wall – 13:57
So they’re saying this is an opportunity for brands to be doing more relevant texting and make that as a better channel. But I don’t know — I don’t know enough about — I know that Community, as a text provider, has some workarounds for some of that stuff that can do it, but now it’ll be weird that it’s baked in. How will that change the tools, and what will the toolkit look like? So that’ll be one — of course, it has to show up first. We’ll see if it actually is delivered and then go from there and pick that up. Yeah, I just had to take a second, too. We want to thank NetSuite by Oracle for their support of Marketing over Coffee.
John Wall – 14:31
For all of our clients, there comes a point where they get large enough and they’re managing so many systems that you’re just caught up in the bureaucracy of it all. You’re actually spending more, maintaining all this complexity. Smart businesses reduce costs and headaches when they get large enough by graduating to NetSuite by Oracle. NetSuite is the number one cloud financial system, bringing accounting, financial management, inventory, HR into one platform and one source of truth. With NetSuite, you reduce IT costs because NetSuite lives in the cloud with no hardware required, accessed from anywhere. You cut the cost of maintaining multiple systems because you’ve got one unified business management suite. You improve efficiency by bringing all your major business processes into one platform, slashing manual tasks and errors. Over 37,000 companies have already made the move, so do the math.
John Wall – 15:17
See how you’ll profit with NetSuite. Again, we’ve seen it firsthand for our clients. Instead of building all these integrations or running batch reports so that you can get inventory and the financials in order, along with the marketing and sales stuff, just get it all in one platform. And of course, having it in the cloud makes a whole slew of headaches go away. By popular demand, NetSuite has extended its one-of-a-kind flexible financing program for a few more weeks. Head to netsuite.com/coffee. That’s netsuite.com/coffee. Again, netsuite.com/coffee. And we thank NetSuite by Oracle for their support of the show. I did want to get in there — you put out a new, updated _Unofficial Guide to the LinkedIn Algorithm_. What’s going on over there? What are the changes in the latest round with that?
Christopher Penn – 16:05
So that is something our friend Amber Naslund, who’s a sales leader at LinkedIn, frequently posts that people ask her and other LinkedIn employees — people ask her, LinkedIn employees, “Why isn’t my post showing up in the feed?” Or, “How does this work? What are the secret hacks that employees get?” And she’s like, “None of that’s real.” And one thing I find myself pointing out in the comments an awful lot is LinkedIn does not make a secret of how it works. You just have to do some detective work to put all the pieces together. The same is true, by the way, of Apple. People would say, “Oh, Apple’s got all this brand new AI stuff. No one had any idea they were working on.” No, if you read Apple’s Machine Learning Research hub, they’ve been publishing about what they’ve been working on for years.
Christopher Penn – 16:50
Some of the stuff that was announced at WWDC this week is in papers two years ago — you know, looking at something on screen and how you compress language models to fit on a device, they’ve been giving this information away. It’s just buried in tech stuff. And so with LinkedIn, it has a fabulous engineering blog. LinkedIn has execs like Tim Yerke who go out and do interviews on podcasts and stuff. And LinkedIn has dozens of academic papers on various services. “Here’s the paper we’re publishing on this, that, or the other thing.” Well, it turns out if you were to gather up all that information, they each contain pieces of how LinkedIn works, how the newsfeed works, how things get — how the decision process happens that allows a post to be seen or not. It’s, like, I think 13 different components.
So there is no such thing as *the* LinkedIn algorithm, it’s 13 different services together that take a data stream and process it. And if you, as we did — if we took all that content, we stuffed it inside Google Gemini, the language model, and said, “Okay, we’ve got all this stuff, now turn this into a coherent single work. What happens when you post a piece of content on LinkedIn?” “Well, step one, you hit post. Step two, follow feed eats it. Step three, it’s scored. Step four is then passed to the trust and safety systems. It’s — it’s embedded, it’s vectorized, it’s passed different storage nodes. There’s proactive and reactive classifiers.” This is all the stuff that happens. And then finally, at the very end of this thing, it spits out, like, “Yes, your post is going to show up.”
Christopher Penn – 18:31
We took all that, we said, “Okay, now that we know what technologies are involved in the order of events, now let’s write a set of three guides — beginning, intermediate, and advanced — for what marketers can do to essentially follow directions. Like, don’t put garbage in your posts and things, don’t put stuff that are low-engagement features in your posts.” And so now we have this unofficial LinkedIn algorithm guide, that’s about a 20-page ebook that has step-by-step — if you’re a beginning LinkedIn marketer — here’s what to do to improve the likelihood that your content’s going to show up. And then there’s an intermediate, and then there’s an advanced one. It’s all in the same guide.
John Wall – 19:15
Alright, that’s cool. I’ll have a link to that in the show notes because people can go grab that. There’s no charge for that, just go pick it up off the site and get rolling. That’s easy enough to do. I’ve got a gear watch. I had my water bottle die — I had this Contigo water bottle I used forever — and I got hooked on this Owala water bottle. So I wanted to actually give it a plug because it’s great in that it has a sealable top, it has both a straw and the ability to pour from the nozzle. So that’s kind of unique because I love — you know, when I’m driving, I want to have a straw so I don’t have to put the canister in front of my face as I’m driving, and — but the ability to pour too is great.
John Wall – 19:52
And yeah, I just solved a bunch of headaches that I had on that. So I’ve got that on there. And then the other one is running apps, too. I would talk about my running stack every once in a while. I’m using an app called WorkOutDoors, which is totally configurable and has a million things in it. I was on Adidas Running, and long story short, the API has changed and broken. Now I’ve lost a bunch of functionality with that. It’s pushed me to try some new stuff. I’ll take a couple more miles on that and then report back on how it goes, but so far that’s pretty promising. As far as — yeah, even — it’s simple toggle of ducking versus muting.
John Wall – 20:26
When you’re getting feedback on how many miles you’ve gone or how fast you’re going, to be able to say, “No, I want you to pause the music” or my book — I don’t want my audiobook running in the background when I get a 10-minute speech on my pace, that — that’s not helping. So, yeah, that’s my grumpy old man plug there for that. Alright, that sounds good. We’ve got a bunch of other stuff coming up. I have Spark Together, SparkToro’s event is coming up in the fall, so I have not been to Seattle in forever, so I’m thinking about — maybe that’s a reason for me to go out there. But we’ll have to see if we can squeeze that in on the calendar. Have you got any — what’s next on the travel log for you?
Christopher Penn – 21:02
There’s a bunch of stuff where we are — you know, from the final stages, getting paperwork together and things. The next thing for sure, yeah, will be in Cleveland, the Marketing AI Conference, a couple of months, but I know there’s other stuff that is pending right now.
John Wall – 21:15
Alright, sounds good. That’ll wrap it up for this week, so until next week, enjoy the coffee.
Christopher Penn – 21:18
Enjoy the coffee.
Speaker 1 – 21:20
You’ve been listening to Marketing over Coffee. Christopher Penn blogs at christopherspenn.com. Read more from John J. Wall at jw51.com. The Marketing over Coffee theme song is called Melo G by Funk Masters, and you can find it at Music Alley from Mevio or follow the link in our show notes.
The post The Future of Onboard AI and The World’s Best Water Bottle appeared first on Marketing Over Coffee Marketing Podcast.
384 afleveringen
Manage episode 423513653 series 2733
In this Marketing Over Coffee:
Learn about what came out of WWDC, how to market better on LinkedIn and more!
Brought to you by our sponsors: Wix Studio and NetSuite
Apple WWDC
Entering the age of on device AI
7:04 Wix Studio is the web platform that gives agencies and enterprises the end-to-end efficiency to design, develop and deliver exactly the way they want to!
iPhone mirroring
Apple has such great UX, why is Apple Music on the Mac not there? But they are paying artists…
Transitioning from SK Ad Kit to Ad Attribution Kit
iOS 18 to add RCS – Rich Communications Text Services
14:28 NetSuite is the number one cloud financial system, bringing accounting, financial management, inventory, HR, into ONE platform, and ONE source of truth.
Updated Unofficial Guide to LinkedIn Algorithm
Gear Watch – Owala, the new best in class Water Bottle, WorkOutDoors for Running
SparkTogether coming up in Seattle
Gen AI Course Updates done: Special Discount on the newest Generative AI for Marketing Course! Hands on excercises to put AI to work for you! USE CODE MOC now!
Join John, Chris and Katie on threads, or on LinkedIn: Chris, John, and Katie
Sign up for the Marketing Over Coffee Newsletter to get early access!
Our theme song is Mellow G by Fonkmasters.
Machine-Generated Transcript
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.
Speaker 1 – 00:07
This is Marketing over Coffee with Christopher Penn and John Wall.
John Wall – 00:14
Good morning. Welcome to Marketing over Coffee. I’m John Wall.
Christopher Penn – 00:17
I’m Christopher Penn.
John Wall – 00:19
And we have all kinds of stuff this week. It’s been crazy, there’s a bunch of things happening. The biggest one we’ll lead off with is Apple’s WWDC, their Worldwide Developers Conference. They roll out some new stuff, but then there’s a bunch of stuff on AI and all kinds of stuff going. So what’s — what was the biggest thing for you? What were you like, “Oh my God, this changes everything?”
Christopher Penn – 00:37
So it’s interesting. They had the big keynote, right? And everyone’s like, “Oh awesome, flashy stuff” like that. But that’s not where the details are. The details are actually in the session because it’s a dev conference, it’s a developer conference. It’s not for the general public. I mean, a lot of people tune into the big keynote to see the stuff, but all the — the really cool/very subtle things are in the individual sessions. The big one to me — and this is very nerdy — the big one to me is the way Apple has architected its AI, which they have cleverly branded Apple intelligence in a couple of different ways. Most of it, from the system architecture diagrams, looks like it runs on the device itself.
So we’ve been seeing, really since like iPhone 13, the number of neural cores in the neural engine on iPhones and Macs and stuff has been getting bigger and bigger. In fact, the new M4 chips that just came out for the new iPads, the previous M3 had the — the neural engine was able to do 18 trillion operations per second. The next generation, the M4 chip, the baseline is 38 trillion operations per second. So they have effectively doubled — more than doubled — the amount of processing these things can do that allows Apple’s custom models to run entirely on the device. So all the cool stuff they’re showing — summarize this document, make a silly image and stuff — all that’s going to run locally, which means that your data is not leaving your device, it’s not going to some unknown cloud provider.
Christopher Penn – 02:10
They announced the integration with ChatGPT, which I thought was very interesting because the way they implemented it is super clunky on purpose. Every time you want to use ChatGPT, you have to confirm like, “Yeah, you’re about to send data to a third party, do you really want to do that?” Which I think is so smart because, A, it allows them to easily swap out another provider — and they’ve said Google Gemini support will be coming at some point — and B, it reminds people, “Yeah, when you send your data to OpenAI or whatever, it’s in most contexts, it’s not private.” So I thought that was really cool.
What was interesting in the Core ML session is they were showing you directions on how you can load your own — how you can convert your models from floating point 16 tensors to Core ML, which is their architecture, and they showed how you prepare it and all this stuff. And then said, as part of your app, you can load this model into the phone or the device and run it locally on device. When you think about it, as long as someone’s really good at model making, that’s huge, because such a big part of the cost of AI these days for a lot of these companies is you’ve got to pay the OpenAI tax or the Google tax, or whatever, to send your stuff to their hardware to process it and come back.
Christopher Penn – 03:33
If your model could fit inside of Core ML, if you can be really efficient and smart about it and make a very dedicated-purpose model — you pay nothing in technology fees because you’re just using the existing Apple hardware.
John Wall – 03:46
Yeah. And we’ve talked about this, even like you said, back when 13 came out, we had said that the idea with this is that it could run these models locally. But, yeah, it’s just kind of crazy now that it’s here. I mean, so is this — the idea is you have — it’s nodes. I mean, you have machine learning running at every single point out in the field, distributed like you said, it doesn’t have to phone home at all. And just a side note too, I did — Elon Musk, went crazy and said that if ChatGPT is baked in, he’s not going to allow Apple devices in the factory or anywhere.
John Wall – 04:19
And the article made clear the point that — well, obviously he missed the idea, which is that it has that button that screams at you saying, “You’re sending this data to a third party,” which, yeah, that is really interesting too, kind of sowing seeds of discontent there. You’re basically telling the world that — think about what you do before you push that button because it’s leaving the device.
Christopher Penn – 04:38
Yep. Generally speaking, Elon Musk is such a petty sore loser that anything that’s a good idea that he craps on is probably going to do really well. So like, it’s like the — I guess, the opposite indicator. If he loves something, like, yeah, that’s garbage.
John Wall – 04:56
You’ve got to stay away from it.
Christopher Penn – 04:57
Yeah, I mean, to be fair, when you look at the design of the Cybertrucks, and you’re like — and they look at like an Apple device, like, okay, one of these things is not like the other. And I thought this was really interesting about the market opportunity for companies with on-device AI. Again, if you are good at building or — you know someone who’s good at building a small, very finely tuned single-purpose model, and you can cram that into — you know, maybe a one billion parameter model, maybe a 150 million parameter model, maybe you have a model that is just trained on your podcast and nothing else. It can’t even tell you what color the sky is, but boy does it know your podcast.
Christopher Penn – 05:40
You would integrate that model into your app in the App Store and have conversational AI in your app that is specifically focused, that gives someone the generative AI experience. I can talk to this podcast app and get, “What is Marketing over Coffee done with mobile marketing?” And because it knows the index, and you have the data and stuff like that, it will do a great job of that. If you ask, “What do you know about the In-Ear Insights podcast?” “I have no idea. I’ve never heard of that podcast and I’m not going to go look because that’s not what I’m trained on.”
This gives people the ability to create rich AI experiences that are baked into the app, baked into hard — run on the hardware, need no Internet other than the data that’s there, and could do really well. So if you have at a shopping guide, and you’re like, “Maybe this guide really knows women’s blouses and nothing else.” You could have a really rich shopping experience using generative AI that’s so tightly focused and never needs the Internet. And so I think there’s a reasonable possibility that an app developer skilled with making AI models will be able to create a huge billion-dollar company app for the iPhone ecosystem that has a server AI cost of $0.
John Wall – 07:00
Alright, yeah, there’s a bunch of other stuff too. We’ve got to peel out of this too, but before we do, we just want to take a second. We want to thank Wix Studio for their support of Marketing over Coffee. I’ve only got one minute to tell you about Wix Studio, the web platform for agencies and enterprises. Whether you manage 10 sites or 1,000, here are a few things you can do from start to finish in a minute or less on studio: set up native marketing integrations in a click; reuse templates, widgets, and sections across sites;
John Wall – 07:26
create a seamless handover by adding tutorials, guides, and more to client dashboards; work on the same canvas at the same time with all your team members; and leverage best-in-class SEO defaults like server-side rendering and automated structured data markup across all your Wix sites. Time’s up, but the list keeps going. Step into Wix Studio and see for yourself. Check it out over at wixstudio.com and we thank them for their support of the show. iPhone mirroring was one thing that I saw, that’s something that — on the Android side is not a concern. But, yeah, that was very cool as far as — especially, you made a point about marketers that have phone apps now being able to do it on the desktop.
John Wall – 08:03
I mean, that is game-changing as far as making headaches go away and being able to type and cut and paste and all that kind of stuff.
Christopher Penn – 08:09
Oh, yeah. I was blown away by that, where you have the virtual iPhone on the desktop, and you drag a file from — from your desktop into the social media management app on the iPhone, as it was — as though it was a local file. Stuff like, okay, that is — because phone mirroring has existed for a while. You’ve had the ability to mirror phones, but it’s always just like, “Okay, there’s the screen of your phone on the desktop, but now you’ve got to pick up your phone and use it.” This is making it a true virtual environment where you can control every aspect of that device from the desktop, and it does not have, like, the crazy lag of, VNC or — you know, the many other things. Such a cool —
Christopher Penn – 08:49
Here’s the thing about Apple, and I’ve been saying this a lot lately, they are not first to anything. They did not have the first personal computer. They did not have the first smartwatch. They don’t have the first smartphone. They did not have the first anything. They have the best. They spend so much time obsessing on the customer experience and making sure that it is flawless and polished that even though there are other market entrants, they simply do a better job having a smooth experience so that someone like our parents — there’s no friction. Like, you don’t have to explain to a parent or a young child, like, “Here’s how you do this, and you’ve got to do the step and open this DLL file and all this.” No, it’s like, “Okay, just follow the on-screen instructions and it will do it.”
Christopher Penn – 09:33
I mean, you’ve — you’ve had that experience with relatives.
John Wall – 09:37
Oh my God, yeah. I have transitioned everybody to Apple just because — to make it all go away. It’s a walled garden that can — can hold them well. But so this does — I had this further down in the notes — but this brings up a sore point of why is Apple Music this redheaded stepchild of this whole thing? The desktop version of Apple Music is just missing so many features and just doesn’t — has so many things going wrong. It’s weird — it’s a cord enough — but I had — over on Threads, we were kicking this around a lot, and there was a wise observation that somebody made that, like, look, Apple Music is just there to sell more phones. Like, it doesn’t compete with Spotify or Tidal or any of those things.
John Wall – 10:13
It’s just that they know that’s one more piece of lock-in to sell more phones. And when you’re selling billions of dollars of phones, then that’s fine. So yeah, I’ll throw that out to the crowd. If anybody else has Spotify versus Apple Music opinions, I’m sure everybody does, we’ll hear that. But yeah, I took it for a test run for a week, and after a week I was like, “Well, I still like the Spotify interface better.”
Christopher Penn – 10:35
Marcus Brownlee did an interview with Tim Cook, which is, like, the first time, I think, a lot of the social media influencers have been doing CEO interviews. I saw Justine Ezarik did one as well. And Tim Cook says something very interesting, which is obvious and true, but you don’t think about it in today’s environment. He said Apple’s business model is not contingent on making you spend more time on your device. Apple’s business model is to sell these devices, and — and yes, the services that go along with them, but it is not to make you an addict to your device the way that Instagram, like, literally has to make you an addict for their business model to work. The more time you spend scrolling, the more time you spend consuming ads, the more money they make. Apple doesn’t do that.
Christopher Penn – 11:12
And so for those services, yeah, it’s not — you don’t need to spend as much time, but also, Spotify is kind of a one-trick pony. It does one thing, which is play music and audio, whereas Apple does many things. So yeah, every individual app probably is not going to be as perfected as a company whose sole focus is that thing. It took Apple, what, five years to get a decent map app on their phone? Right now, Apple Maps is good, it’s very good, but it — you know, they ceded a lot of territory initially to Google Maps because the original Apple Maps sucked.
John Wall – 11:45
Yeah. And then if you want to go full audio geek, Apple has a bunch of stuff in Atmos, so if you’re — you know, top-end audio stuff, you can get that. But that opens up a whole ‘nother audio file door of, just because it’s a new format, does it sound better, yes or no? And it really all comes down to individual listening, like, you have to just go listen to all these tracks and figure out which one is better. Unfortunately, there’s no easy solution as far — and the equalizer is the other one that goes in there. Spotify does have an equalizer even on the phone app, which iPhone doesn’t have yet, but they do have on the desktop.
Christopher Penn – 12:19
Yeah. The other big thing for marketers is they’re moving on from SKAdNetwork to their new ad attribution kit. So this is new software within the iPhone, the iOS ecosystem, to help advertisers understand better how the ads are working and things. But it’s very interesting, all the data is aggregated, so you get no user-level data whatsoever in any way. So this makes it compliant, obviously, with all of the major regulatory requirements for advertising. But it’s also interesting the way they positioned it in the tech session on it. You get less data, the smaller the audience is that you’re advertising to. You get more data as your audience gets bigger because there’s less chance of being able to re-identify someone from the ad data you get. So they were showing off, as your audience gets to really big sizes, yeah.
Christopher Penn – 13:08
Now you can have more things like gender and age and stuff like that, whereas you have a very small audience, like, “Nope, you had people, but you had people download your app.”
John Wall – 13:16
Yeah, they blocked that off. That makes sense because that is so easy to game. You can just start to individually tag people, that makes sense. Oh, yeah. Another interesting thing on the tracking side too, is with iOS 18, they’re going to be supporting RCS — this Rich Communication Services — which Android has had forever and Google has put forth as a text message set of standards. And so the idea is now we won’t have this weird gap between you’re sending GIFs and emoji responses that are tied to the bubble and all that kind of stuff via Apple that you couldn’t do on the other side. And they’re saying that there’s actually more tracking data on that too, you’ll actually get analytics as far as what’s happening with text.
John Wall – 13:57
So they’re saying this is an opportunity for brands to be doing more relevant texting and make that as a better channel. But I don’t know — I don’t know enough about — I know that Community, as a text provider, has some workarounds for some of that stuff that can do it, but now it’ll be weird that it’s baked in. How will that change the tools, and what will the toolkit look like? So that’ll be one — of course, it has to show up first. We’ll see if it actually is delivered and then go from there and pick that up. Yeah, I just had to take a second, too. We want to thank NetSuite by Oracle for their support of Marketing over Coffee.
John Wall – 14:31
For all of our clients, there comes a point where they get large enough and they’re managing so many systems that you’re just caught up in the bureaucracy of it all. You’re actually spending more, maintaining all this complexity. Smart businesses reduce costs and headaches when they get large enough by graduating to NetSuite by Oracle. NetSuite is the number one cloud financial system, bringing accounting, financial management, inventory, HR into one platform and one source of truth. With NetSuite, you reduce IT costs because NetSuite lives in the cloud with no hardware required, accessed from anywhere. You cut the cost of maintaining multiple systems because you’ve got one unified business management suite. You improve efficiency by bringing all your major business processes into one platform, slashing manual tasks and errors. Over 37,000 companies have already made the move, so do the math.
John Wall – 15:17
See how you’ll profit with NetSuite. Again, we’ve seen it firsthand for our clients. Instead of building all these integrations or running batch reports so that you can get inventory and the financials in order, along with the marketing and sales stuff, just get it all in one platform. And of course, having it in the cloud makes a whole slew of headaches go away. By popular demand, NetSuite has extended its one-of-a-kind flexible financing program for a few more weeks. Head to netsuite.com/coffee. That’s netsuite.com/coffee. Again, netsuite.com/coffee. And we thank NetSuite by Oracle for their support of the show. I did want to get in there — you put out a new, updated _Unofficial Guide to the LinkedIn Algorithm_. What’s going on over there? What are the changes in the latest round with that?
Christopher Penn – 16:05
So that is something our friend Amber Naslund, who’s a sales leader at LinkedIn, frequently posts that people ask her and other LinkedIn employees — people ask her, LinkedIn employees, “Why isn’t my post showing up in the feed?” Or, “How does this work? What are the secret hacks that employees get?” And she’s like, “None of that’s real.” And one thing I find myself pointing out in the comments an awful lot is LinkedIn does not make a secret of how it works. You just have to do some detective work to put all the pieces together. The same is true, by the way, of Apple. People would say, “Oh, Apple’s got all this brand new AI stuff. No one had any idea they were working on.” No, if you read Apple’s Machine Learning Research hub, they’ve been publishing about what they’ve been working on for years.
Christopher Penn – 16:50
Some of the stuff that was announced at WWDC this week is in papers two years ago — you know, looking at something on screen and how you compress language models to fit on a device, they’ve been giving this information away. It’s just buried in tech stuff. And so with LinkedIn, it has a fabulous engineering blog. LinkedIn has execs like Tim Yerke who go out and do interviews on podcasts and stuff. And LinkedIn has dozens of academic papers on various services. “Here’s the paper we’re publishing on this, that, or the other thing.” Well, it turns out if you were to gather up all that information, they each contain pieces of how LinkedIn works, how the newsfeed works, how things get — how the decision process happens that allows a post to be seen or not. It’s, like, I think 13 different components.
So there is no such thing as *the* LinkedIn algorithm, it’s 13 different services together that take a data stream and process it. And if you, as we did — if we took all that content, we stuffed it inside Google Gemini, the language model, and said, “Okay, we’ve got all this stuff, now turn this into a coherent single work. What happens when you post a piece of content on LinkedIn?” “Well, step one, you hit post. Step two, follow feed eats it. Step three, it’s scored. Step four is then passed to the trust and safety systems. It’s — it’s embedded, it’s vectorized, it’s passed different storage nodes. There’s proactive and reactive classifiers.” This is all the stuff that happens. And then finally, at the very end of this thing, it spits out, like, “Yes, your post is going to show up.”
Christopher Penn – 18:31
We took all that, we said, “Okay, now that we know what technologies are involved in the order of events, now let’s write a set of three guides — beginning, intermediate, and advanced — for what marketers can do to essentially follow directions. Like, don’t put garbage in your posts and things, don’t put stuff that are low-engagement features in your posts.” And so now we have this unofficial LinkedIn algorithm guide, that’s about a 20-page ebook that has step-by-step — if you’re a beginning LinkedIn marketer — here’s what to do to improve the likelihood that your content’s going to show up. And then there’s an intermediate, and then there’s an advanced one. It’s all in the same guide.
John Wall – 19:15
Alright, that’s cool. I’ll have a link to that in the show notes because people can go grab that. There’s no charge for that, just go pick it up off the site and get rolling. That’s easy enough to do. I’ve got a gear watch. I had my water bottle die — I had this Contigo water bottle I used forever — and I got hooked on this Owala water bottle. So I wanted to actually give it a plug because it’s great in that it has a sealable top, it has both a straw and the ability to pour from the nozzle. So that’s kind of unique because I love — you know, when I’m driving, I want to have a straw so I don’t have to put the canister in front of my face as I’m driving, and — but the ability to pour too is great.
John Wall – 19:52
And yeah, I just solved a bunch of headaches that I had on that. So I’ve got that on there. And then the other one is running apps, too. I would talk about my running stack every once in a while. I’m using an app called WorkOutDoors, which is totally configurable and has a million things in it. I was on Adidas Running, and long story short, the API has changed and broken. Now I’ve lost a bunch of functionality with that. It’s pushed me to try some new stuff. I’ll take a couple more miles on that and then report back on how it goes, but so far that’s pretty promising. As far as — yeah, even — it’s simple toggle of ducking versus muting.
John Wall – 20:26
When you’re getting feedback on how many miles you’ve gone or how fast you’re going, to be able to say, “No, I want you to pause the music” or my book — I don’t want my audiobook running in the background when I get a 10-minute speech on my pace, that — that’s not helping. So, yeah, that’s my grumpy old man plug there for that. Alright, that sounds good. We’ve got a bunch of other stuff coming up. I have Spark Together, SparkToro’s event is coming up in the fall, so I have not been to Seattle in forever, so I’m thinking about — maybe that’s a reason for me to go out there. But we’ll have to see if we can squeeze that in on the calendar. Have you got any — what’s next on the travel log for you?
Christopher Penn – 21:02
There’s a bunch of stuff where we are — you know, from the final stages, getting paperwork together and things. The next thing for sure, yeah, will be in Cleveland, the Marketing AI Conference, a couple of months, but I know there’s other stuff that is pending right now.
John Wall – 21:15
Alright, sounds good. That’ll wrap it up for this week, so until next week, enjoy the coffee.
Christopher Penn – 21:18
Enjoy the coffee.
Speaker 1 – 21:20
You’ve been listening to Marketing over Coffee. Christopher Penn blogs at christopherspenn.com. Read more from John J. Wall at jw51.com. The Marketing over Coffee theme song is called Melo G by Funk Masters, and you can find it at Music Alley from Mevio or follow the link in our show notes.
The post The Future of Onboard AI and The World’s Best Water Bottle appeared first on Marketing Over Coffee Marketing Podcast.
384 afleveringen
Toate episoadele
×Welkom op Player FM!
Player FM scant het web op podcasts van hoge kwaliteit waarvan u nu kunt genieten. Het is de beste podcast-app en werkt op Android, iPhone en internet. Aanmelden om abonnementen op verschillende apparaten te synchroniseren.