Get in touch

These days, artificial intelligence is being sold as the solution to all of your technological problems. But is spending time and energy on AI systems worth it? According to Paul and Rich — it’s not. They share why it’s more important to clean up your existing problems instead of implementing new, sparkly technologies to solve them.

Transcript

Rich Ziade The term Einstein became appropriated by like high school kids like “Oh, great move Einstein!” [Rich laughs] It’s not even good anymore!

Paul Ford Yeah, “Thanks for predicting my Q4 sales results, Einstein…”

RZ Yeah, “Well done Einstein…” [music ramps up, plays alone for 15 seconds, fades out]

PF Richard.

RZ Paul, you know, we love to call out bullshit on this podcast.

PF Ah, do we?

RZ We pulled back a little bit because we’re like, hmm, we got big client names now and we’re actually very successful and we got to watch ourselves. But we’ve applied some discipline but you know what I’m gonna go back in the well here one more time here and just call, I want to call bullshit.

PF Okay.

RZ You know, we often look at the big players in our world and that’s, that’s usually driven by—

PF Your Facebook’s, Google’s, all those guys, right?

RZ No, no, but also in our world world. Like you know the guys like the big consulting firms and then and then you’ve got you know, Gartner and Forrester coming out with their with their, you know, “Hey guys! Internet of Things!”

PF I love—you know, Gartner has made billions of dollars from two lines, they just draw that like that little—

RZ The X-Y axis?

PF Cross line. It’s the world’s dumbest Tic Tac Toe game. And you just sort of like “Yeah, alright, fine.”

RZ Exactly. And and I want to touch on something that has been and continues to be a big part of how technologists and decision makers feel like they need to be on top of, and it turns out, it’s mostly bullshit.

PF Wait, wait, wait, I don’t even know you’re gonna you’re gonna spring this on me. What subjects are you talking about? 

RZ I’m talking about AI and machine learning. [Paul laughs] I’m not gonna name names. But we’ve lifted the hood on some big, big platforms at Postlight. We get called into some big places. And they say you should see our AI engine and tell us if we’re doing it right. And it turns out, it’s just regular expressions. [Rich laughs]

PF Oh, yeah. If you want to charge more for statistics, just call it machine learning. Yeah, it’s—absolutely!

RZ Now, here’s what I’m getting at. There is such a thing as machine learning, there are some pretty wild things, you can make a cat look like, you know, Joe Piscopo. You can do all kinds of interesting things with machine learning.

PF Frankly, once that’s done—[Rich laughs]

RZ Here’s what I’m getting at. There is a pressure that exists for many businesses, businesses that aren’t in technology. I’ll give you an example. A big retailer, has a CIO, and probably has a CTO. And that retailer is constantly thinking about how to stay ahead. So let’s, let’s give an example. Let’s pick—Target—Target’s technology is huge. It’s a huge undertaking, there’s always Amazon breathing down your neck, you got to be on top of your shit, right? So Target has a CIO, CTO, I don’t know who this person

Chief Information Officer, Chief Technology Officer. Okay.

RZ These are people whose job it is to make sure that the tech is up to snuff. And there is the baseline which is don’t get hacked. Target went through an awful hack a few years ago, absolutely devastating.

PF Oh, it’s you can literally just say the ‘Target Hack’.

RZ The Target Hack—it was a bad scene. So as a baseline, your Chief Security Officer and your CIO and you know, you got to make sure the thing is up and running, and not hacked and fast. And works on all the phones, even old shitty phones, right? That’s the baseline. But then there’s how—there’s that what I call the ‘How do I get ahead?’ mindset, right. And then what you have are a lot of whispers. And they happen at conferences. They happen by Gartner and Forrester, they happen through product lines of big platforms. And they say you got “This is the AI component you can add on.”

PF So this is every big new, every big platform now so your you know, your CRM systems or your RP or whatever, instead of saying, “Hey, we’re gonna make this really easy and fast to use. And we’re gonna also make a charting module that’s super simple. So you’ll be able to get your work done and export it and then you’ll go home at the end of the day.” Yeah, they don’t say that. They say “We’re gonna cloud enable you with machine learning services.”

RZ “That’s going to be predicting what you’re going to do next, knows what your customers are gonna want to do.” Look, I’m dispelling all this as if it’s actually a sham. It’s not a sham. But let me tell you what else is not a sham. Clippy was not a sham. 

PF I’m so glad it was Clippy like for a minute. I’m like, I know you really well, but I was like, Whoa, this might be it. The rug might come out from under the whole damn thing.

RZ For those that are that don’t know what this is, or too young, Clippy was this helper that came in and when you’re using Microsoft Office, and it would come in it would bounce for a second. If you did a formula bad in Excel, it would say “Hey!”

PF Everybody knows Clippy because he shows up in memes, which is how the young communicate.

RZ That’s right. So I guess what I’m getting at is this. It isn’t all a sham as to whether it is actual intelligence or the artificial kind, let’s keep our pants on here. It’s not intelligence.

PF AI as a concept is just utterly cultural garbage. It doesn’t have anything to do with technology. It’s just like, it’s machine learning and prediction and neural networks and so on. And I’m gonna tell you, I’m good at learning really boring stuff. Woof, boy, when you get in on this one, as you go further and further into this world, it’s like man, like, “You know, the reinforcing network, blah, blah, blah” which is why those people are making $900,000 a year because the kind of brain that gets excited about machine learning is like 10 times worse than the kind of brain that gets excited about blockchain, like just that already is just—

RZ Yeah, and the point I’m trying to make is not to shit on this stuff. There’s actual it’s very early days, people don’t realize it’s a lot to do. There’s a ton to do. It’s not nothing isn’t actually intelligent yet. It’s a lot of brute force computing. But what I want to get at is this, that “Hey, CIO, CTO—”

PF “Jeff, Hey Jeff, and other Jeff.”

RZ Diane, there’s some Diane’s out there.

PF No, it’s Jeff.

RZ Here’s the thing, you really want to leap ahead. Just clean up your thing, the amount of opportunity that sits on platforms, and I’m talking about platforms that are worth billions. 

PF Rich, just save your breath, I love you for saying this. It’s nice. Just why wander around this particular wilderness? Because here’s, here’s what’s happening while you’re out—we’re on our podcast. Hey, everybody is great. It’s great. A couple 1000 views, we love you write us an email. Hello@postlight.com. Here’s where you are, you are in the wilderness, you are in the desert going “AI is a lie! There is one true God and it’s user experience!” Okay? You know who isn’t in the wilderness with you? Deloitte. Deloitte is sitting there giant consulting firm with an army of consultants, and they’re saying we are going to use 360 enabled Salesforce AI to predict—

RZ Einstein. It’s called Einstein.

PF Yeah, we’re gonna use Einstein. That’s right, as a brilliant branding move calling it Einstein. It’s like calling a dog Fido. So we’re, we’re gonna predict exactly how many you know, cat collars, you’re gonna want to sell, you’re gonna want a manufacturer. In Q7, you’re like, well, there isn’t a Q7, they’re like we’re still working that part out, right? That’s, that’s where they are with this. So it’s a tidal wave and your CIO and your CTO. All they know, they can’t fix anything at the scale that they’re at a Target. You can’t I don’t think people understand this, you can become the CTO at Target, you can’t fix anything. You can delegate to one of five direct reports that you’d like to see some better KPIs, you’d like them to set some metrics for other people that other people can use to set metrics. 

RZ We love to shift from venting and complaining to real advice. Want to give a good piece of advice to CTO or CIO, especially a CIO. I mean, you know, often doesn’t come from them. It often comes from the CEO. And what is this. I want my damn search results to be less than 10 seconds. Why is it taking 45 seconds? I’ve got 500 engineers, why is it taking 45 seconds for search results to come up?

PF First of all, 10 seconds is garbage. It should be like so it should be like a millisecond. But regardless.

RZ I don’t know what it’s rummaging through, it’s not Google, like it’s a proprietary platform. Whatever it’s doing. It’s like our customers are complaining. Why is it taking a minute for this page to load? Why is it taking 25 seconds for search results come in? 

PF Needs more Einstein.

RZ Exactly. This is this is it, right? People think that new tech will fix the old. It’s not the case. 

PF Yeah, it just makes it worse! 

RZ There is nothing. nothing expires worse than rotten milk than old software. Software gets rickety and old. And it’s not fun and there’s nothing to sell. It’s just clean up. Make it fast.

PF You know what this is? This is exactly the same as I remember talking to somebody, I went to a little school upstate, and it’s real easy to get people to give you money to build a building with their name on it. Okay, it’s real hard to get money to pay for the janitor in that building. Just famous problem. Nobody wants the Paul Ford Memorial Janitorial Service Closet. They want the Paul Ford Memorial House of Computers, right? So for whatever and I’m sure if you’re out there go like, nah, man, I would never do that. Well, for whatever reason, when people get a lot of money, they like their name on things. So put that aside for a sec. I feel that in some ways, this is the—corporate leadership has the same damn problem, which is I want to do the ML-enabled-super-future-proof-exciting-mega-project and then you come in and go “yeah but the basement is actually filled with live fish. The leak is so bad, that there’s like carp in there, we got to get the carp out. And then we got it like dig with a shovel!” And they’re like—

RZ Where’s the failing there? Where’s the failing? Because everyone wants the shiny new thing. So I can’t, I can’t entirely blame the technologists and the CIOs here. Like how are you not turning to people and “We’re losing customers, because our salespeople can’t demo the thing because it takes like a minute to load.” It’s embarrassing.

PF It’s bravery, man. It’s bravery. Because what you have to do is literally turn your whole org and say, “This is all bad, and you have to do better. And there’s no way out except for going in and fixing what’s wrong,” and then they’re gonna turn to you. And they’re gonna say, “Well, we can’t because we invested in, you know, technology x and technology x lasts us—the company that made technology x is gone, and the replacement is going to cost $4 million.” And now you’re going to go spend $4 million on the same damn thing. And that’s awful. Or you could plug-in Einstein. [Paul laughs] It’s like a little animated guy with the—looks like a bobble head. And he will predict your Q4 sales revenue and, woof, boy, that’s that’s some good stuff right there. But you know what? The CEO, you know, where the CEO used to be? Used to be the CFO, right? CEOs doesn’t know what the hell you’re talking about. So you say to the CEO, “I need $4 million to get the carp out of the basement.” And they’re like, “Why are there carpet in the basement?”

RZ Yeah, you know, I think there’s a couple of things happening here. One is the CEOs feel like they have to be somewhat informed on the tech, they have to be right? They have to ask the right questions and be informed so they can click it. They gotta sign off on that budget. 

PF Yeah, I know, but they sign off on that Gartner quadrant, man.

RZ That’s right. That’s right. And so which is very forward looking.

PF You know what’s hurtful? Like I love, you know, not love, but I have tremendous respect at some level—and fear—for Google, like Google is I both fear and respect Google. It’s a hell of an engineering org. You know, if you go to like Google websites for some of their services, like cloud services, they mentioned their Gartner quadrant. And that, to me, that’s like saying, for Google to say that we’re in this Gartner quadrant is like me saying that my third grade teacher thought I was a good writer. Like, it just feels like—

RZ It’s outerspace. [Paul laughs]

PF It’s very humbling, like, just like, but they’re like, “Yeah, you know what? We got to do it. We got to tell them, the Gartner says we’re good.” And it’s like dude, you’re Google, I use Google Docs every day. And you got to tell, you know, a carpeting company CTO, that you really belong in that top right quadrant? And hell yeah, you do. Because that’s the misbegotten IT services, enterprise shit world we live in. Everybody is just swallowing that all the time. And then really, you and I, and Postlight shows up, kinda when it’s just, they can’t take anymore, right? Like Deloitte didn’t fix it.

RZ A lot of our opportunity lies in wading into the the problems, right wading into the challenges that an org is facing. We are strange in this way. First off, we’re very value driven, not time driven. So our goal isn’t to go and just infect your organization and stay there forever, and never get out. Like our goal is to actually deliver value not bill time. That’s, that’s part of our model. 

PF We’re gonna pull the wisdom teeth out, not make you lie there and be like, “Hey, what if we leave them in for a little bit longer, but just add some new teeth on top?”

RZ Yeah, exactly, exactly. And that’s a cultural thing. And it’s something that, you know, is sort of born out of kind of our approach to things which isn’t, we don’t know what Gartner said, this year, or last year, or the year before that. We don’t know. Like, we’re very pragmatic about software.

PF You know, it’s funny, though, we’re on the list now. I get calls from Gartner on a regular basis. They want to get us in there. And I don’t understand—I’ve talked to them. I’m like, alright, well, let’s see what Gartner has to say. I can’t understand a word. I can’t understand. “You know, organizations of your size.” And I’m like, okay, what size is that? Like, “your scale, for the things you offer…” And I’m like, what is that? Right, like, they don’t care. You know, though, let me just throw a curveball in here, too. It’s not just about selling the AI. What I do love is when they come to us with sunk costs, and they’re like, “Can you fix this?” This is often the Postlight, the critical Postlight moment is they come in, they’re like, “We spent $12 million.”

RZ It’s not good. 

PF “And we’re out of money.” And we’re like—and it’s always an awkward first conversation, cuz you’re like, you have to pay me money for services too, like, I can’t fix that. Like, ’cause there’s, there’s actually a fundamental fallacy, which is like, “Is there any way that you could do this for free, ’cause I spent all the money, but it was so much money. I gotta have something to show for it.” But it doesn’t work. Oddly enough, that often is the beginning of the conversation.

RZ It often gets to that, you know, I think for a lot of organizations, they know that a lot of it is some cost. And some of it is a way to here and there like that’s part of the process in a way. But I think this is driven by a few things. And I ultimately I think the technologists on the floor are actually partly responsible for two reasons. One, technologists love to play with new shit. Okay, if you’re telling them they get to mess with the new thing, and they got to go into a new space, a new domain, they love the idea of it. So that’s attractive though, so they’re not gonna shun that. Two, technologists are very defensive when it comes to you telling them that their child isn’t performing well enough, they gave birth to the thing, right? Not only gave birth, they’ve been nurturing that thing and fixing it and refactoring it here and there for years, and all of a sudden you’re coming to them. And this is why the non technologist is so powerful in their signal, because they’re coming in and saying, I hear you, I know, it’s hard. I know, we have a lot of SKUs, but I need it to load in less than 10 seconds. And it’s 45 right now!

PF Yeah, look, I mean, it’s not just that you’re saying, I don’t like your child, you’re actually saying, “We’re gonna decommission your child and get a new child from a large cloud provider.”

RZ Yeah! And what they’d rather do is say, it’s often written off as like, technologies now moving forward, we must keep moving forward with it. It’s like no, wait a second. There’s nothing cool about cleaning up the thing. Right, there’s nothing fun about it.

PF We talked about this on a previous podcast, right, which is just like the thing that we do, we write five times as much, we provide probably five times as much functionality, especially like the API level, per, I don’t know, per week or per month, of programmer time than we used to, because what’s happened in our industry is that everything has gotten so commoditized and standardized. And an example would be rest API’s or graph QL APIs is backed by a database that don’t need to scale to 10s of millions, but might need to service like a million or 2 million people a year, that domain is pretty well established, it actually it shouldn’t take a lot of time to get something bootstrapped, get something usable, if you know what your data looks like, you should be able to get. slap something in front of your database and go to town pretty quick. But that idea that the thing that A) the thing you used to do, is actually the right thing, and is still the thing that you should be doing today. Don’t worry about AI and ml and all these new things. Yeah, get that. B) that should be really simple and fast. And you should use a more fluid framework to do it. C) Once you get all that right, and you get those queries down to one or two seconds, and you know, max for the really big and complicated ones. Now, you can start to say, “Hey, what can we do with this data? You know, everybody’s excited about AI and ML, maybe I’ll take that seminar” Right? But that’s six to eight months to bootstrap the more modern thing that works pretty much like the old one, but just doesn’t suck is bad. If time nobody wants to spend.

RZ Not only is it time nobody wants to spend, nobody else has said “this is what success is for you.” What they’re saying—-

PF Yeah, CEOs got to do it.

RZ CEOs got to, I think it comes from outside of, we do it because a lot of times our secret sauce is essentially wearing the hat of the business stakeholder and putting pressure on the tech side, and cutting through bullshit, right? That’s a lot of what we do.

PF Literally right before this, I was teaching a 90 minute seminar, I’m going to do three installments about how API’s work and what platform dynamics can do for a particular organization. 

RZ And your audience was non technologists.

PF Just literally a class with slides, you know, with 12 people in attendance, so that they could have a collective understanding of, you know, the point I’m making is that you can have more and more transactions, and that that the cost goes down with more transactions instead of up. And that’s pretty exciting. That’s, you know, just trying to get those ideas across. But it’s so much easier to lump it all up and say, “Boy, if you don’t know what’s happening over here in ML, if you don’t know what’s happening, machine learning and artificial intelligence, you’re gonna get left behind it is 1999, as far as you can tell.” And that’s good consulting. When you get them that anxious. And then you come in you say, “Don’t worry, I’m gonna, look at this. I got it. I got an arrow in a box. You’re in luck.” 

RZ I think the other thing I’d add is this isn’t about an either or, I mean, you can have your R&D group thinking about sort of this modern stuff and making sure my you know, Amazon Echo can actually order me sunflower seeds, a bag of sunflower seeds. That’s all great. Like, I think you can still continue to do that stuff. But what we see time and time again, is how much neglect there is around the core thing because it functions, it’s like an old boiler. Like it’s just, it’s heating the building. leave it be. Meanwhile, it makes this banging sound and every so often it bursts into flames and you got to just take the fire extinguisher out. So it’s not good!

PF There’s 18 inches of sludge.

RZ Yeah! You just don’t want to deal, so what you’d rather do is actually put in the sound system in the walls. You’d rather spend the money there. 

PF You know, just to completely torture and ruin this metaphor. It’s not the boiler switching out the boilers. Okay, it’s finding the good plumber, then he’s gonna tell you more bad news. Like, what if you get ripped off? Yeah, you know, if you have a good plumber, you’re kind of like okay, I guess it’s time for that new. You’re really out of money. But most people who own a house are probably, you know, and are going to have no, no that that cost is coming. 

RZ You know, I gotta say shame on our fellow consultants, especially the really big firms that A) sell all kinds of bloat and be, it’s shit, they actually deliver shit, they deliver more of the same shit.

PF They don’t know what it is either. They don’t program, it’s like systems architects who draw a lot of boxes, they didn’t hack together some horrible website that went down under load, because everybody, you know, then everybody yelled at them on Twitter, like they’re not motivated by that. They’re motivated, they’ve been taught and then go to seminars. And you know, they’re part of some division that is some company that got bought nine years ago, and is now the strategic solutions provider for you know, scalable internet architectures and multi cloud tenant platforms. And you’re just sort of like, you know, when you can’t even list the division of the company, when you can’t see the name of it in less than a minute, then how are you even going to get a solution?

RZ No, you’re right. And the truth is, it all comes back to metrics, it comes down to the non technologists, you could say it to a consulting firm, and you could say it to your CIO and CTO, “I need it to be less than three seconds, period. Like show that to me on a consistent basis.” 

PF Honestly, just say it, let them freak out.

RZ Yeah, let them freak out. And then let them push back and say, well, it’s impossible. Not impossible. The pet store website across the street is less than two seconds.

PF Well, you know what they’re gonna—what I love then—is that the answer is gonna be like, “Well, they use the Shopify API, look.” And you’re gonna go, “Why don’t we use the Shopify API?” It’s like—

RZ Yeah, exactly, it just goes round and round.

PF Yeah. “Our Virtual Private Cloud won’t allow us to use any external services.” Well why is that? 

RZ Yeah, yeah. And on and on.

PF And then, you know, by the time you get to just pure acronyms, you know, you’re starting to get close to the wire. So it’s just like, what do they say HIPA. Yeah, where they say, if anybody says 59s, just whatever you got, value you’re getting, you’re getting down to the metal.

RZ I think this is fundamentally the piece of advice, which is if you’re a non technologist, right, you often get bullied. And you’ll often get obfuscated language to make it confusing and complicated—

PF And you’re being bullied by non-bullies. That’s the worst. If you’re the CEO, you’re probably more likely to be the bully in this dynamic. But so, you’re literally, it’s all the kids that used to throw stuff at high school. [Rich laughs] It’s horrible.

RZ The truth is, you’re allowed to say, “Well, I want it. I don’t care. Why not? What do you mean? It’s very complicated? What do you mean it’s gonna be 10 million bucks? Why, why? Why?” And you’re allowed to say that.

PF Don’t do it in the moment, though, ’cause they will acronym you back and you’ll look like an idiot, because you won’t know what they’re talking about. What you get is “I want to know why person X can do it in less than a second. And I want to plan, I don’t care what you have to do. I don’t care if we have to start a new business. I want to plan for how we get there with this data.” Nerds under threat spew acronyms, it’s almost like a like a, like an octopus sprays ink. Just like you come at them. And they just stay this cloud the air like an octopus. But if you come to them, and you say “I actually, like you know, you’ve been trying to hold this thing together. But get me to a second per query. I think you can do it, come back in a week with a plan. ‘Cause I know you’re the guy. You’re the people who have this. You’re the ones who can do this. Nobody else can do and somebody else tells me it’s impossible. But I think you’re the one who’s actually not gonna bullshit me. They’ll come back with a plan.” 

RZ No, I think that’s right. I think that’s less adversarial. And I think going back full circle to AI and ML. It’s like, how did we move on to the shiny new thing, when the fundamentals are busted? And I gotta tell you, Paul—

PF I mean, it was blockchain. It was blockchain 18 months ago. Am I on ‘blockchain meet’ with you right now, am I recording this on my blockchain enabled audio program? No, I’m not, what a surprise because that’s what I was promised.

RZ To close it out, I think it’s worth highlighting that we go into some very, very big organizations. And it’s incredible how much neglect we see in certain corners while they’re looking at the shiny new thing. And that’s real.

PF I love when they reach out though, and they’re just like, I know, this is probably the worst you’ve ever seen. [Paul laughs]

RZ Yeah, yeah. The honesty. You do appreciate the honesty.

PF Yeah. But it’s like, no, trust me, it’s actually as bad where you are as it is anywhere else. 

RZ That’s right. That’s right. 

PF We should tell people about Postlight’s new AI platform. 

RZ Yeah, exactly. Look, you know, this sounded like one long ad when you know, it’s not meant as such. It’s just It’s just what we’re observing out in the world.

PF It is. It is, let’s be clear.

RZ It is. But you know, a lot of times, look the conversations, that’s where it ends, and we don’t, they don’t become clients of ours. But we actually pride ourselves in giving like kind of plain English advice about stuff.

PF This is not, it’s not an ad because this is the conversation we have with people all the time. This is a conversation simulator. We didn’t give you any room to talk as a listener. So that’s pretty accurate. [Rich laughs] No, I mean, if there’s a point to take from this, right, it’s that everybody actually has a surprisingly normal range of problems. And it’s spackling wonderful new technologies on top of it. It’s not a solution. It’s just deferred, it’s not even deferred maintenance. It’s just more spackle, it’s another thing you have to take care of.

RZ And it becomes costly and costlier to take care of it. I mean, that’s, that’s the thing. Ultimately, this is about money.

PF Somebody asked me today, while I was teaching my API seminar, everything starts to rot, right? Like I keep reading about all these things that cost $50 million to fix and don’t work. How do you do that? You know, how do you keep that from happening? And I’m like, you create as little as possible, you raise a little bit and you focus, you really focus on getting your data, right, and then putting as light a layer in front of that data to manage as you possibly can, and let the database do all the work, let the cloud platforms do all the work. And you should be doing as little as possible, and you should be doing less each year. But still doing it, you should never turn your back on an API because it will jump out of the water like a shark and bite you.

RZ Check us out, Postlight.com, we’ve got some great case studies in there. [music fades in] And what you’ll see in them is a good amount of what we talked about meaning pragmatism, good strategic thinking, great execution. Don’t be afraid to hold others accountable. Put success metrics in front of people, regular real world success metrics, not “I must have this tech inside the walls” it’s more like this needs to be really fast and really easy to use. Like those are things you’re allowed to say to this day.

PF Be proud of your ignorance of implementation details, and but also spend more and more time learning about what is—there are really well understood metrics for good experience. That’s what the CEOs should know. Right? It should be fast and usable and fluid and all those things. And then you ask about those don’t ask about and even seconds are like too long, right? Like it’s so yeah, go in there, figure that part out and then put your foot down, make better experiences, please! AI is not gonna fix it.

RZ I’m gonna go talk to my Amazon Echo and Google Home and Siri. 

PF Absolutely. Alright, we’re gonna do it, we’re gonna figure this all out. Check us out at Postlight.com and we’ll talk to you next week.

RZ Have a great week.

PF Bye! [music ramps up, plays along for 3 seconds, ends]