Connected Commerce: AI Gives You Speed and Depth. Humans Make the Decision.

Hai Mag 00:08

Welcome everyone and welcome to the Connected Commerce podcast. And I'm your host today, as always, Hi Mac, the founder and CEO of Eva Commerce. And it's my distinct pleasure today. I have Ruben Ali Kanyan together with us. And Ruben and I, we know each other for a couple of years now. He is in that Amazon e-commerce space, he's one of the veterans.

Hai Mag 00:35

Welcome, Ruben.

Ruben A 00:37

hi. hi. Hey, it's great to be great to see you. It's been it's been a while as we started a conversation. you know, that that that we that we we knew each other and we met. So it's great to be here and thank you for the invitation. Great to be here.

Hai Mag 00:49

Absolutely. So today the kind of the format of this podcast is talk about connected commerce stories, like how you connect two things together so that you know it works even better. So that's kind

Ruben A 01:04

Yeah.

Hai Mag 01:04

of the idea. Let's start with your first story, Robert.

Ruben A 01:07

yeah, so for that I I think I'll speak a little bit about the what is a lot of sellers, service providers, and anybody who is in any way, shape or form involved with an e-commerce, the AI connections to the e-commerce, and how we utilize it within within our company. And utilization for that will be based both on

Ruben A 01:31

How an agency we use to improve our systems and operations internally, and how we also use that to improve the brands that we oversee and their performances.

Ruben A 01:43

the uses of it, you know, the most basic one came in with utilizing it within our communication and just trying to streamline anything that would take off a workload from either it's the brand director or an account manager, senior account manager, anything that we reduce them. But you know obviously as it grew with especially more and more coming on and entropics and the cloud, we just started getting

Ruben A 02:07

deeper and deeper and to the point where we we started having in weekly meetings j just on AI specifically, understanding where we are, what can we do, and without exaggeration, to the point that the person ahead of that management was specifically requested not to come up with anything new with our management because we've been going so fast where the team could not even catch up with all the uses

Ruben A 02:31

of the AI specifically with Cloud. We trusted everything to Cloud that we've been that we've been utilizing and we've been kind of putting in front of them. And that went from having basic reporting tools or some automations to the point where right now we have a full access to every brand out there and can simply interact with our agents

Ruben A 02:55

Understanding every account where they are and using multiple tools that are also connected to get ins and outs in the bottom of every situation that is out there. Find a solution based on using data dive tools and then the management into implementation. Not to note that also with it it it improved the ways that we do the reporting and an automation of it with the various dex creations.

Ruben A 03:22

like our partners we work with, they're just praising you know, the the depth of reporting that they get and the frequency around it. So the use of it being extensive and and being powerful, but again, the way we were approaching it, we're not doing it for the sake of doing it and and and and working because we're using an AI. And then the way I started this conversation, like I asked to stop, it it meant one thing because

Ruben A 03:47

the opportunities become start becoming infinite what layers we can add, but the value to that will start diminishing as well if it's not properly used and it's not implemented properly by the team. So y absolutely

Hai Mag 04:01

Very interesting. Tell us, you know, like

Hai Mag 04:03

it's very, very interesting, by the way. sorry to interrupt, but I think that, you know, one of the things that you mentioned, even the AI is growing exponentially, but human brain, human mind is linear and it's the adoption, it's it's just a matter of adoption of the human

Ruben A 04:19

Apsoluty, yes.

Hai Mag 04:21

rather than the AI. AI is already far away. It's better to

Ruben A 04:25

Yes.

Hai Mag 04:26

stop sometimes innovation and wait for the catch up. And also, of course, there is still hallucinations, there's still some guardrails you need to put the into that AI implementation. Can you give us some specific examples? Like you mentioned dashboards, reporting, some automation, but I mean like maybe something specific that you now leverage AI and use it.

Ruben A 04:48

Yeah, I'll I'll yeah

Ruben A 04:51

I'll I'll speak to something that I think a lot of you know will will be interested which is first of all execution it is not trusted to AI whatsoever. So right now the use of AI is speed and depth. Those are the two areas that we are mostly focusing on. Speed

Ruben A 05:09

Because we can get an information faster than before. And in some cases, right now, the way we're experimenting, ahead of something's happening. So it's either ahead of something's happening, we're getting an information as a warning sign, or something happened, we've noticed, we're identifying the core issue to be able to act. We utilize the speed of that to understand, and then we can utilize the historical data of.

Ruben A 05:33

You know, asking, for example, suggestions based on the data, but never trust the execution. None of the tools we use are designed to execute anything for us. And I was a big advocate from a day one, especially when a lot of people don't realize in Amazon space AI automations to the degree being far ahead than anything else within beat optimization tools and all that advertising optimization tools available.

Ruben A 05:59

But I always, the one thing that I hated among all these platforms available, not naming any names, was managers trusting those automations for the tools to be done without them intervening with that and making the decisions behind it. So the same concept is right now, but in an exponential level with everything else, from understanding where the ranking went, you know, where the traffic is going, where the traffic is coming. So speed.

Ruben A 06:24

Is one thing the way we use it, faster identifying where new issues. And even though we can ask suggestions to understand what it can be, which we'll get again based on actual data, the decision making and pulling the trigger, it stands still behind the managers and assigning the strategies for an advertising or making decisions on the creatives or making decisions on the SEO. So that is the biggest use that that we get

Ruben A 06:48

That is, you know, it it it's simplified the life of everybody around the organization.

Hai Mag 06:53

Ruben, do you think that I you mentioned about speed and death? And I fully agree with that, because we do the same thing now. It's very fast that we know there is an issue on the account, or we can get so many different types of tools connected to AI so we can get an analysis of the situation. Now, do you think that it is a matter of time?

Hai Mag 07:17

for AI also to come up with the execution or a part of the execution, or you don't think that it will ever happen.

Ruben A 07:26

Good good question. I have recorded a video, only it's it's already on YouTube. If it's not, it's gonna be about my predictions about an accelerate, the Amazon accelerator. I want to start with

Hai Mag 07:35

Okay, okay.

Ruben A 07:36

that because one of my predictions is Amazon is going to push their own AI agents, which are the event essentially the executioners of the things that you know right now you know the entire

Ruben A 07:48

Amazon seller signal dashboard is different the way you interact and it all goes to that. So if I'm going to guess, I think that option will be absolutely 100% there in the near future. As I started my conversation, we do have AI agents already in our systems as well, in our Slack, where we can absolutely tell and ask for them to do certain things. Again.

Ruben A 08:11

Will that certain thing to be as higher level execution on a problem that we see, or it will be gather data on, let's say, a very interesting example, the search query performance for the brands. Certain brands want it. We track a search query performance. We specifically track on 20 top keywords our week over week market share ownership. It's a very important method.

Ruben A 08:36

If you're doing something specific ranking campaigns, the brand owners want to know how we're doing specifically on those keywords as a market share. So, can we ask an AI agent to prepare that type of report and share with the client on a regular basis or share with us to review? Yeah, absolutely, yes. I I think you know we we're gonna get there. We are already there. And in terms of a larger execution, I will still be skeptical at this point to trust completely a management, but it's no doubt.

Ruben A 09:03

That there's gonna be companies in the near future who's gonna come with a program or with a software or a solution that will tell you this is your VA right now, and you can trust him to do everything for you. So that VA will analyze it, will tell you exactly what needs to be done, and will do it for you. It is, it is, it's gonna happen. How much you wanna yeah,

Hai Mag 09:24

Yeah, it's something like yeah, go ahead.

Ruben A 09:27

how much you wanna trust to that VA.

Ruben A 09:30

to do things for you, it will be up to the brand owners, service providers whatsoever. But it's gonna be happen but but it will have.

Hai Mag 09:34

Isn't it

Hai Mag 09:35

a bit like this autonomous driving started six years ago when I was in Palo Alto and that was like maybe ten or twenty cabs like Venmo cars, and now San Francisco is full of them. It's almost even hard to find an Uber driver. And in Austin now the robot taxis from Tesla are almost occupying the space. Every week they are putting more

Hai Mag 09:59

Robotaxis, which are probably replacing real Uber drivers or taxi drivers, but it happens slowly because if it happens right away, then it will also create some other problems with the number of unemployment and everything. So everything is happening in some pace and it will

Ruben A 10:16

Yeah.

Hai Mag 10:16

happen. But one of the things that you said about Amazon agents, for example, I I wouldn't

Hai Mag 10:23

I mean, I do have a brand too. I wouldn't trust, you know, even if I was not in agency business, I wouldn't trust Amazon agent for a very simple couple of facts. First of all, I'm not gonna tell my my cocks to Amazon, like I wouldn't do that. Secondly, Amazon is a marketplace, they sh cannot favor one to another brand. Where now the interesting fact with AI development is it's not about the data that you know.

Hai Mag 10:50

Like maybe the reason of you selling more is because your competitor's price increased or he's out of stock. And you know,

Ruben A 10:58

Exactly. Yeah. Yeah.

Hai Mag 11:00

like one of the things that I'm seeing it as a poker game now in the AI rather than a kind of a backgammon game or a chess game. It's not a chess game because your competitors, for example, if somebody goes out of stock, your system thinks, I'm doing great.

Hai Mag 11:16

The agent thinks your agent thinks or even by the way, our brand manager sometimes think, I'm doing great. Let's reduce the bits a bit because we are doing great. It's a doing ACOS is going up. But the real reason is the competitor, you know, a price or an out of stock

Ruben A 11:30

Yeah. Yeah.

Hai Mag 11:32

situation. And typically in that type of a poker game, I say poker because you don't know exactly all the parameters of the

Ruben A 11:40

Yeah.

Hai Mag 11:41

other side. I mean it

Hai Mag 11:43

becomes really complex and you need

Ruben A 11:45

Yeah.

Hai Mag 11:46

that human decision making like for example

Ruben A 11:48

Yes.

Hai Mag 11:49

in a poker game you kind of go for the patterns and try to understand but still you know it's very difficult. but if it is Amazon, they have all the data, okay, they can play the game. But if they play the game with that agent, it w it won't be like a I mean

Hai Mag 12:05

I think even legally it won't be l the right game. So that's why probably agencies need to build that agency or leverage that agency in the future.

Ruben A 12:14

no

Ruben A 12:15

yes, and and I'll give you another good example of the inefficiency of Amazon or trusting everything with Amazon agents. I and before I say that, I still think that Amazon agents or that development within the Amazon in some segments will have superiority against everybody else and I'll specifically name calculating your profitability will have much better superiority against any best tool out there.

Ruben A 12:41

And I'm part of a company, B Profit. I'm on a board of advisory there. And when we haven't our our entire platform is to calculate profit, but most of the things I absolutely agree. Not only you cannot trust Amazon, I think that you know they are not gonna be you know as as good in performing those tax. And and and the greatest example is here with their inventory calculations, right?

Ruben A 13:06

So

Hai Mag 13:06

Right.

Ruben A 13:07

if you if you have a product that sells 4,000, 6,000 units on a monthly basis, and then let's say circumstantially something happened and you got out of a stock for 30 days, right? So the Amazon recommendation is going to be a couple of hundred units for you. Because in a 30-day segment, it's not calculating all the previous months because they have the limit of giving you the advice of what to reorder.

Ruben A 13:32

and they're not gonna be accurate with that. you know, will that with the time, you know, get again get better? And and then Amazon will start, you know, getting more nuanced into certain aspects. Yes, but I think it's just a way bigger game, and it's a lot longer when Amazon in all aspects of it or the agents aspects of it will be as good. So that's why the internal agents for the agencies are way better value right now.

Ruben A 13:58

because you not only collect the data, you know the day-to-day, you've also had the insights to the brands that the Amazon is not going to have. So your understanding and then agents, AI agents' use of understanding is going to be a lot more deeper in recognizing the patterns and making the decisions based on those patterns than Amazon will. And I don't think Amazon is striving to fix all those issues right now, anyway. They know that they're

Ruben A 14:22

start at something that will be bigger as they usually do, but they're learning.

Hai Mag 14:26

And Ruben, maybe two more things. One is the brand and the agency is on the brand's side, meaning that like you know, because

Ruben A 14:32

Absolutely, yeah.

Hai Mag 14:34

you don't have any other like if the agency wants to get paid, I mean you need to grow the business and you need to bring the profit. And so it's on the same side. You're the advocate. but Amazon is not. Like there's different you know, kind of strategies or whatever is important for Amazon is not the most important for the brand.

Hai Mag 14:52

The second thing is, for example, you know, TikTok shop is now very popular for the last specifically nine, ten, nine months. it's very popular. Okay, it's a discovery engine, Chat GPT, AI, discovery engine. And maybe as a brand, now you start understanding that the discovery is not really happening on Amazon. Still people are, you know, kind of starting on Amazon when they have a strong intention.

Hai Mag 15:20

But it doesn't

Ruben A 15:20

Yes.

Hai Mag 15:21

mean that the journey started on Amazon, meaning that

Ruben A 15:23

At an Amazon, exactly.

Hai Mag 15:25

the agents that you are developing, and that's why I call it the connected commerce podcast, because I don't believe personally on Amazon alone. What I believe is what is the end-to-end customer journey with all these AI components here and there, social commerce popping up and AI popping up, all of them together.

Hai Mag 15:48

Maybe Amazon is where people like to finalize the transaction and Amazon always wants to go to the awareness, but I don't

Ruben A 15:56

Yes.

Hai Mag 15:57

think that game is an Amazon game. As we speak right now, maybe it may change in the future. But right now I believe it's more of a a chat GPT game, a perplexity game, a cloud game, because that's what I do, right? I personally do.

Ruben A 16:11

Yeah.

Hai Mag 16:12

what do you think about that?

Ruben A 16:15

no, I'm gonna agree with that. I'll and I'll give you another example of what you're saying. It's the customer path to purchase with the DSP.

Hai Mag 16:23

Right. Right.

Ruben A 16:25

we're calculating for the brands who do that, it's an average of five touch points before the purchase happens. you know, it it it circles. So, you know, th there there is a huge validity with that. specifically with the relationship of the

Ruben A 16:39

All the AI components available right now, you know, ChatGPT. I know they are not participating in any of that from an awareness standpoint, because they have locked their claws in them. All right. You know, they don't want to be discovered, they want to grow their their own Alexa for for shopping, but that doesn't really mean that the future discoveries that will happen in cloud will not or or Chat GPT or product discoveries will not end up to your point back to the Amazon.

Ruben A 17:06

even though there are no links or direct channels to go to Amazon through those you know through those discovery engines at this point. for the the TikTok, TikTok is an in interesting which is they they they're significantly behind on this race, let's let's put it that way, but the the signs that it's going to start supporting is is is definitely there.

Ruben A 17:28

First of all, from discovery standpoint, TikTok is not does not have the limit reach, meaning you can find products through the AI that is connected to the to the TikTok. But more interestingly, I see that some of the TikTok supportive tools like Color Data are starting using TikTok to support the it essentially operates as an AI agent for the

Ruben A 17:52

you know, for the TikTok sellers by identifying anything about the product, asking suggestions, typing. So it does not have its native platform to support for the sellers, but definitely side companies are working towards it to be able to you know deliver that as i'm gonna be value component no doubt

Hai Mag 18:10

Very interesting. I agree. There is one more thing that I think that legally, very soon we're gonna see if there is a real influencer or if it's an AI influencer, very soon we're

Ruben A 18:21

yeah.

Hai Mag 18:22

gonna see like, this is AI generated video. And now there is a a big conf confusion or a conflict, right? Because if you see this super nice, beautiful lady and very happy to put whatever cream on her face, but it's an AI generated video.

Hai Mag 18:38

Video, what's

Ruben A 18:38

Yeah.

Hai Mag 18:39

the point? Like, I mean, if I am a buyer, why I'm like watching an AI generated person recommending me to buy this. Of

Ruben A 18:46

Yeah.

Hai Mag 18:47

course, it's good if it's how to type of a video, how to use this product. It doesn't matter, it can be AI. But if I'm really looking for an influencer, I want a human influencer

Ruben A 18:58

Yeah.

Hai Mag 18:58

because people buy from people. That's why you

Ruben A 19:01

Absolutely.

Hai Mag 19:02

and I still

Hai Mag 19:03

in the business because l you know, people want to know like who who am I going

Ruben A 19:07

No down.

Hai Mag 19:07

to trust when my ASIN is suspended? Not to an agent, you know, because that agent doesn't fix your problem. So

Ruben A 19:10

Yes, not that yeah. little listen, there are a lot of issues

Ruben A 19:16

with that, you know. Again, it's a couple of things that I spoke very very much so was the character seventy-five character change with the titles. I've

Hai Mag 19:25

Right.

Ruben A 19:25

been talking about and warning like, hey, if you are not changing, Amazon's gonna change it for you. And when Amazon

Hai Mag 19:30

Right.

Ruben A 19:31

changes for you, he's gonna change it based on the algorithmic command it has.

Ruben A 19:36

Is not gonna care about you know certain aspects of your brand and it can put you within the suspension risk without you even recognizing it. Because it may use some material based on your back end and front end and construct and title that will be not in a compliance. Now you have a problem in your hand. So you can't trust that. also on the same subject, I don't know if you caught that, but Amazon had put a specific rule, which is a you know good one if you if you're regarding regarding your last conversation was

Ruben A 20:03

a use of AI imagery. There is now a specific rule if you're using AI imagery, and especially when it's the kind of an influencer, when you're using a field humans, you have to mention that it's an AI generated. You cannot have it you know without it. Otherwise it will be violation, a strict violation. As a matter of fact, that comes from Amazon itself being sued in New York. So

Hai Mag 20:24

That's right.

Ruben A 20:24

yeah they they you know they they they they know the the risk of it.

Ruben A 20:28

And you mentioned something that was interesting. you know, everybody has their views about an Amazon, but to me, Amazon is not really a seller's friend. They they they definitely pretend to be the one, but all their pretendness comes from one thing because they're a friend for the buyer. So when they pretending to be your friend, they're mostly pretending it for you to be better.

Ruben A 20:50

So the customer can benefit more. So that is an i important kind of an subject to know for the for the sellers to realize that you know whatever is out there that Amazon is recommending is is not out there for you much, but for their customers. So you know, having that in mind and making the adjustments, it's it's it's a good call.

Hai Mag 21:08

Absolutely. Look, Robin, it was an amazing conversation about all this AI agents, connected commerce with them, and then you know, all the implications. I think it's very exciting to see the development of everything in the next couple of years. I think will be definitely transformational. So you run an agency

Hai Mag 21:30

But you didn't tell us like, you know, how people can reach out to you and how they can, you know, search for your agency. Tell us a bit more.

Ruben A 21:38

yes, absolutely. We're you know I'm available twenty-four seven at PASAgency dot com. I'm very active on LinkedIn, so you'll feel free anybody who has any questions, reach out to me directly. Those messages come to me and I'm the one checking it. So I'll I'll make sure to get back and any social media out there. I'll be I'll be it'll be great to hear from you guys. And there's a lot of questions about anything AI and we're gonna have a lot more. Sellers gonna have a lot more questions after this Amazon accelerate.

Ruben A 22:05

probably even more after Amazon unboxed you know later later this month and i post my regular contents on the youtube so if you want to follow there i'm gonna put every update there but if there will be any confusion around it how to use it how to get better at this feel free to reach out and shoot me dm maybe to connect

Hai Mag 22:21

Thank you very much, Ruben, for our for being our guest. And hope to see you soon at Amazon Accelerate in Seattle. Thank you.

Ruben A 22:29

Yeah, absolutely. Thank you for having me.