#272 The Future of AI in Retail and E-commerce: Insights from Connectly's Co-founder, Yandong Liu

Transcript generated by Podium.page
Help us spread the word by tweeting about us at @podiumdotpage and including us in your shownotes! https://podium.page

NOTE: There were 2 speakers identified in this transcript. Podium recommends using "Find and Replace" to change the speaker label to the appropriate name. Speaker separation errors can arise when multiple speakers speak simultaneously.

0:00:02 - Mehmet
Hello and welcome back to a new episode of the CTO show with Mehmet. Today I'm very pleased to have with me Yandong Yandong, thank you very much for being with me on the show today. You've got a very rich background and I don't like to steal the lights from my guests, so I would keep it to you to introduce yourself. 

0:00:42 - Yandong
 Thank you for having me, mehmet, on your show. It's an honor. My name is Yandong Liu. I'm the CTO and the co-founder of Connectly. I guess we can talk about Connectly a little bit later. 

First, quickly about myself. I was born and raised in China. Then I studied computer science in college. I grew an interest in computer science and AI fairly early on. After college I came to the States to pursue a PhD in computer science at Carnegie Mellon University. I was doing all kinds of research. I was actually publishing, but more on the theoretical side. After graduation I moved to Silicon Valley, like many people did. I started working at various tech companies. For example, I first worked as a researcher in the Yahoo research, sort of continuing my academia life, publishing all that. Then, after Yahoo, I went to Uber in 2014, while Uber was really growing through some explosive growth. After Uber, I became CTO of Strava, the popular fitness app. I was managing everything technical there. Then, after Strava, I started Connectly. 

0:02:09 - Mehmet
Great, great to know about you more, yandong, and what a career, I would say. The question that came to my mind for this journey you started more into the academia and then you shifted to more, let's say, the practical part of it. This is something we don't see much. Honestly. We see usually people who go into the academic career. They stay there and they rarely go out from there. What was the moment that you felt you said, okay, I need to take this now to the real world? What was the motive behind that? 

0:02:50 - Yandong
Yes, that's a very good question. Indeed, I was very interested in more like academia life work. I was planning to become a professor. If I'm not doing this, of course, I shifted my mind many years ago. I was still interested in the theoretical and designing a side of that, especially when it comes to machine learning and AI, but I was more fascinated with the potential of AI of transforming real-world interactions and making a real-world impact. That's why I'm doing this. I'm always looking for real-world application Out of AI. I use messaging every day. When you combine the two ideas, that was pretty exhilarating to me. The concept of starting something new from the ground up, when you have the freedom to innovate and push boundaries. That was all very, very attractive. That's why, when I had this opportunity of starting something new that connected, I decided to just do it. 

0:04:18 - Mehmet
I always enjoy listening to such inspiring stories, yandong, because I was not in the academia in the sense as a professor, but because I was working for a university in the technology department and I had to interact a lot with the professors and sometimes I was telling them you have these really cool ideas. It wouldn't be nice if you take it out. The thing I really liked your story because you stepped out of a comfort zone, as they say, because now we will talk a little bit about the work you do with Connectly, because it needs really a brave heart to go into this world. But what inspired you actually to choose? Of course, you've done a lot of work before, but to take the past of AI power, conversational commerce, why did you choose this specifically? What was the problem that you see or the opportunity that you have identified? 

0:05:22 - Yandong
Yeah, absolutely so. Connectly is all about creating this meaningful one-to-one relationships between business and their customers through the power of messaging. I'm a big fan of messaging. I use WhatsApp, I use iMessage, I use sometimes SMS when everything else is not available. I hope I think you do it too. I mean pretty much that's what I see. 

Everyone loves and uses messaging every day. It's probably safe to say that everyone has one of more chat apps on the phone. We use them every day, right? We use messaging to stay in touch with our family and friends, and the experience has been great. We've been doing more and more things like video calls, audit calls. That's all great, but on the other hand, the potential of business messaging is largely untapped. 

We don't really talk to business via messaging. Whenever we would like to get in touch, we probably still need to make phone calls, email them, go to web chat, sometimes even through form submission, sometimes when we try to get help, which is really not great In my opinion. All those often don't work. You try to get ahold of them on the phone. You have to wait for 10, 20 minutes if you're lucky, and then you have to. Oh, this is how I spell my name. I have to read out my credit cards, I have to repeat myself oh sorry, you're not talking about the billing. Let me transfer you Now. You just repeat your whole story again. All those are not great. So here we see. 

On the other hand, messaging is a far superior experience. It's asynchronous, it's two ways. You don't have to stop whatever you're doing, you just fire and forget. That's great. And companies are building new capabilities every day. You can manage your appointments, you can run payments, all that. 

So here I'm pretty excited about the bridging the gap using messaging and transform this into a powerful and efficient way of for business interactions. So, and also, we're focusing a lot on conversational commerce, like you pointed out. So the idea is to allow people to be able to like, browse the product catalog, get product recommendations, ask all kinds of questions and eventually finalize the payment, finalize the transaction by running payments, to have the full end to end shopping experience without downloading. I don't like the idea of downloading and managing a app for every single brand alike. Probably the merchants think the same way too. Right, they don't want you to do the other additional work, they just want to be able to engage and do business with you. So that's why you know we're pretty excited about this idea the bringing the full shopping experience to messaging and without people doing the extra work. 

0:09:09 - Mehmet
That's really cool and to your point, yeah, especially in the place where I'm living, in Dubai and Middle East. You know, actually we do, because maybe you know you're very familiar with these chatting apps. But people sometimes get surprised. I tell them, you know, believe it or not, we do business transactions and business negotiations over these chatting apps and the reason is because it's convenient. You know, like you have the phone in your pocket, all what you need to do to type and then, as you said, because it's asynchronous, people can wait. You know they don't have it's not like you know, and for some reason people love it more than email For some reason yeah, yeah, no, I can tell you here in the region where I'm living, it's the standard like business really. 

Sometimes it's run fully on WhatsApp. I've seen this or other like chat applications, so 100% on this one. Now, I know you know the question is obvious, but you know I love to hear it from you Now because you have also the you know the factor of AI into the picture. So how this is different and on from, let's say, these bots that we see, they put them in customer support. So how do you differentiate from that perspective? 

0:10:35 - Yandong
Well, first, we focus a lot on like marketing and transactions and, you know, getting product recommendations or that, so we're not doing much customer support today, which is a different business. The second is we do use AI, probably similarly everywhere in this whole messaging experience. What I describe is it's pretty, everyone loves it, but it has to work with. Sometimes. We all got frustrated by the mentioning the chat bot experience and or you build a nice tools and platforms. Now you have to hire, you know, a couple of human agents to help. On the other hand, we like to introduce this new use case and to businesses and their customers, without adding the new work, hopefully we can automate, actually minimize the work. So this is where AI is playing a big role and, for example, I can give you a couple of examples why we think our solution is going to be great for a lot of people. One is marketing. All businesses need to do some marketing work and then often make like marketing campaigns or that for different events, holidays or even promotions right. So traditionally, if you run it for email others, you have to really design everything, generate the copy, find the images, decide who you sent to and all that. So lots of work. Here we use AI to almost fully automate. We have a product called one click campaigns. You just tell us what your goal is. For example, maybe your retailer you're trying to promote I don't know skincare products to your new user, user who recently signed up, right, you just tell us oh, I'm doing this and AI will do the rest. Ai will generate the text, generate the images and adding buttons, because what's up in other similar platforms? They have very rich experiences, the buttons, highly interactive and not all. The beauty of our marketing campaigns is they're not one term but multiple turns. So, for example, the first message oh, we have this great promotion. By the way, everything is personalized, so people find it very relevant. People like this kind of experience. Oh, do you want to know more? Or you want to opt out? If you want to know more, we show more messages and you can have a conversation with our AI. You can get product recommendations, who have different questions before making the purchase decisions. So people really like this kind of decisions. 

We're kind of also creative in this process. We have, for example, sometimes we have like give me five. We have some givenification, which is really kind of unique and interesting compared to this traditional marketing approaches. One example we had is we can give me five the marketing campaigns like. We make one like oh, do you know when I don't know XYZ company was founded. Then we show a couple of options and buttons, right, just a simple tap. If you got the right, it will give you 10% off. I think we really see this as a uniquely way of engaging customers and ROI is great. Compared to email others, roi has been huge, but this is not easy, right? If you do it, you need to do it manually, but with AI, because we scrape, we have lots of data about your business. 

0:14:23 - Mehmet
We know a lot Everything can be automated. 

You know like this is again, it's residing with me as a consumer. You know like I always think about it and, unfortunately, some retailers because I know you focus on retailers and e-commerce so they got it wrong. And you see this. You know they copy paste the same thing. So you're telling me here I will get some personalized aspect, based maybe on some of the information, on some of the transactions that I have done before with you as the merchant, right. So now this is one part, but other than this and you know, because we're talking here about the retail customer experience so, in your opinion, other than you know the few things points you mentioned, what other gaps you think connectedly is also aiming to bridge? 

0:15:20 - Yandong
in addition to this, Well, personalization is a big part of it. Other than that, there is a data gap. We need to know a lot about your business in order to personalize. We all like that experience, but we need to know a lot about your products, your business, your audience, not only to personalize the message, but personalize the cohorts of that. That's one. 

Another challenge, as I see here, is integrations. Our goal is always to make a plug-and-play turnkey solution. You can start your first campaign in five minutes, something like that. But to build the full solution, we need to build lots of integrations into the platform. For example, we need to know where to get your product information. We need to know where to fetch your audience information. Not only will we make those great campaigns and allow people to purchase, but we need to pass the data back. We need to show your great analytics. We need to show past some of the insights back to your system. Today, nothing works in isolation. To make this fully automatic, we need to build lots of integrations Marketing, billing, ticketing, customer support, payments we're doing all of those, yeah. 

0:16:59 - Mehmet
That's great also, yandong, to know about this. This is for anyone in this space, especially retail e-commerce, like you should have a look, I would say, Because I think, yandong, do you see this as that? Something optional anymore for retailers? I think this is they must have, because otherwise they will be left behind. I'm asking you this because, also, I want to see your view how things are going to move in the future. Now, it's like the customer experience, I would say, but where are you seeing this conversational AI taking us in the future? 

0:17:41 - Yandong
It's already happening as we speak, especially from where WhatsApp, or messaging, is already heavy as an important part of life, such as Latin America, southeast Asia, the Middle East. We also have a lot of clients from the Middle East and Europe. I fully agree. I think that businesses should be actively thinking about adding this as one additional growth channel. It's already happening. It's highly efficient. ROI is great compared to some other communication channels. Very effective. I see people not only engaging but actually converting, just finishing purchasing of that Plus. It's very easy to get started. You don't need to develop a new website, new app or have developers of that Plus. Hopefully, most of the work is automated, especially for mid to smaller retailers. If you're large, you might be able to afford building your own app and running your own marketing campaigns, especially those data driven. That's not easy, right? It's not like you can do everything manual. 

You have to use lots of data points to optimize your campaigns, optimize your audiences, all that. But not every business is able to do that, especially the smaller ones. They don't really have any resources or energy to do that. 

Here we are really building a plug and play platform to enable the smaller ones, so that if you're Shopify or some other VTex platform users, you just come to our platform through a few clicks. You're to go to go and it's always on right. We decide through the whole user journey. If you recently signed up, we send you this message. If you're highly engaged, we send you something different, that you feel less active. We send you some messages to re-engage you. Everything is automatic and you don't have to go to the Plus the experience. 

Why experience is better is we all receive marketing emails, like you said. First, they're not personalized, which is very cookie cutter, and the first thing I saw is to unsubscribe. So that's a big difference. Second is, even if, when I'm interested, I click on it, I'm redirected to the website. I log it back in adding the things to the card, I apply the coupon code, I finalize it, which is not really great, especially on the phone and here on WhatsApp. Oh, you saw this, you like it? Oh, one tap, I'm good to go, I'm ready to buy. You know everything about me, you know my mailing address. You probably have a credit card already on the profile and you just need a confirmation from me and to purchase with this number ending with one, two, three, four. You're done right. So I think this is going to be a great experience for everyone. 

0:21:11 - Mehmet
Yeah, like you mentioned a couple of things which are very important actually about. Again, it's back to the personalization and happy that you answered one question I had in mind. So it doesn't matter, you know, even if they are small, because you talked about the data point, but I can understand that you can help them actually, even if they are small, to build the system for them, which is great. Now, one word you kept repeating and I'm happy you are repeating it actually which is automation, because I talk a lot about automation and I tell people you know, of course AI is the buzzword, but in my opinion, the power is in automation. Of course there is a power in AI. I'm not denying this, but I love that you mentioned this and you know how do you see? You know, like, if you want to judge, let's say, on a technology in general, like which is more important to say like it's AI powered or actually it's to say it's automated? 

0:22:14 - Yandong
Yes, automation is the key, I agree. You know we all love great results, but nobody wants to do the work, let's be honest, myself included. So the more automated the we can automate, the better. So here, of course, we automate a lot through AI I kept saying that but also a lot of other, like integrations and workflow management, which is actually very, very important. For example, if you can just set up the work a user journey once, and it's always on, is there doing the work? It's a dream scenario, right? You don't have to come back to our platform every day and make a new marketing campaigns and engage people. It's sort of doing the work automatically. And while any users are enjoying the experience, our work is turning the messages into revenue. That's the goal. For example, I can mention oh, if we did detect that. Oh, you are asking for product recommendations, then our AI module is triggered and then we'll looking for recommendations in the back end, fetch the latest and recommending relevant items to you. Or you're looking for oh, I don't know. 

Sometimes I often after I, get started with a very big idea rather than knowing what I'm looking for. It's not like a one click. It's not always you know what you're doing. You're just looking for to purchase with the best price. Often I don't know what I'm buying. For example, maybe I'm buying a sweater for my wife, but I need a recommendation. I give the bot of AI my budget, the color preferences, all that, and then I get a couple of items. Then I give AI some feedback oh, I like the style, but do you have something more affordable? I got more recommendations, all that, and then I have follow ups. 

Is this machine washable? Very, very common. I do it actually all the time. I look for such detailed information, but only through the fine print. Why good Amazon? Good that I go through every little detail I meet? Sometimes I go to YouTube for a couple of product reviews, but it's very time consuming. I don't necessarily do that every time because I don't have the energy, but our AI pre-scan all such information. So instead of just doing all the work, you can just ask our AI and get all those information answered. So I think this is a great experience. And follow ups right. You can ask AI oh, after a place more, where is my order? The AI knows we need to have a. This is getting a little technical, but AI is able to have an API call to the third party, a backend, fetch all the information and push it back to the client. What's have I message right. So this is all examples of automation. I believe this is the key to success 100%. 

0:25:45 - Mehmet
And a couple of months back I was running these kind of educational series about, especially when the AI started to appear more with chat, gpt and all things I was telling people it's not only about the generative AI itself, it's the power of automating it, and I said, if you don't use automation like, you're losing a lot actually, and just what you mentioned now. So you described it perfectly. What is the pain today for the consumer or for the client? And every one of us live this. And you are bringing the in-store experience. You know when you have the assistant and you can go and talk to them. So I'm bringing this to this world, which I love it. 

And, by the way, one thing also, it's a matter of fun, just to throw some fun here, when you said about the email that comes to you and you unsubscribed, which I think 99% of people they do the funny thing is people which they get it wrong because they didn't implement it the right way. So I start to see these messages and there's a small note at that saying, hey, this message was partially written with AI and it seems that they are using it in a not, I would say, proper way. The message is no one I'm sure had the time to maybe read it even once. Because I will give you an example. I received just today an email such you know. Title is nice, the subject of the email is good, it's catchy, but you know the way they tried to personalize it. 

They relied on an old data of mine, so they were talking about previous place where I used to work, and I said, ok, now they scraped the data from somewhere. It's not OK. So of course I unsubscribed because still, you know, and but I love what you're doing here, because it's something a little bit different. Now I want to switch a little bit gears and talk about the fact that, being a co-founder, and then because I know that recently you have and congratulations on that raised your Series A. So how is this experience to be into this startup ecosystem going? Talk to VCs, investors, so how was this experience from your side? 

0:28:15 - Yandong
Oh man, it's been quite a journey doing a startup. It's quite different from working at a large company, but it's been fun, a lot of learning. Well, fortunately, I've been CTO a couple of times and I've navigated through different phases of growth. I've been through some early stage. I've been through some rapid expansion growth, so that helped me a lot. I was able to bring a lot of experience into the job. That's one. And I've been, fortunately, on both sides of AI, including both the designing and also application building. So that was good. But yeah, as a startup, by definition, we face new challenges every day, right? 

Even when we talk about this great shopping experience, for example, it's not. We didn't encounter any challenges. I'll give you one example. It's data issue. All that this happens every day. It's all very pretty to describe it experience. In reality you have to deal with, for example, the data is missing, the merchant has a million different ways of describing the same size and color. You can imagine, especially work across different merchants. All those small issues combined become big problems. Right, you have to really be able to address those every day. And because the fundraising, all that we all know that it's not exactly easy to raise these days. I'm very glad that we're able to do that because our strong growth, our vision and strong team, all that. So, and hiring I spent a lot of time hiring and building as we become AI. We put together a very strong AI team. 

We build, we hire some of the best talent from the market and we're building with some of the latest technologies, including GPT-4, gpt-4 Turbo from OpenAI, palm 2 from Google, and we're also planning to roll out our own models because we have lots of data. We have a huge amount of unique data where we have four performance feedback right. This is very unique in my opinion. I want to speak a little bit more to it. Lots of people they build AI applications but they don't necessarily have the performance feedback or how the data is used, how much performance the data is bringing or how much help the data is doing for the business. On the other hand, for us, for every single message, we know exactly if the message was opened, read, clicked on, if the people actually purchased something from the message and how much. If they did, how much was the value right. 

We have the full performance feedback. So that will enable us to build a fully tailored model other than those foundation models. Those models are great. We're going to continue tuning them, but I think to have the full experience to fully tune, tailor the experience for our customers, we need to eventually build our own models. So that's some of the challenges we're facing by doing a startup, but I'm enjoying this. I'm building something new that people enjoy using our business growing day over day. I'm very happy with the progress so far. 

0:32:27 - Mehmet
That's great to hear. 

It's good that you mentioned the challenges, because this is one of the things I wanted to ask you and thank you for mentioning this, and it's an eye-opener what you mentioned and this is especially for the techie people who listen to the show about. 

I love the approach you took. So you relied on the available large language models like GPT-4 and Palm and so on, but you are building your own, because this is a question I always ask, especially if I'm talking to CTO or CPO on the show, and even sometimes I ask people who are just from the business side and say how comfortable are you in relying on a third party? At the end of the day? Because there are plenty of applications now that are built on open AI basically, and if open AI API goes down, everything goes down. So it's a good, also eye-opener. Now Yandong has almost where, coming to an end. Now I have also part of the audience, a lot of first-time founders and, I would say, new entrepreneurs or French graduates. So, if you want to, we want to keep with them some wisdom words, what you can tell us. 

0:33:47 - Yandong
It's a very good question, mehmet. So I've been thinking this a lot myself and especially I'm a first-time founder too and, to be honest, I'm learning a lot as we go. But if I had to have to think of one piece of advice, I would say that Now everyone is building something with AI away, but you really need to figure out the one problem, one significant problem, that AI can actually help with. That's the key. We shouldn't be doing AI because everyone is doing AI, or that's the way you think you can raise money, or because the technology is getting so good oh, I have to add it to my product. You really need to identify one problem that AI can help address, and you need to build a winning product to address that need. 

In my opinion, either need to make life easier, minimize work or completely get rid of the work, or 10x the outcome In our case. Actually both right, so one click, or that always on. So this is the key, and then you build a sustainable business, right? But then the only way to achieve, the only way to build a sustainable business is customers find it valuable. That's it. That's my simple recipe. Hope that helps. 

0:35:28 - Mehmet
You're not only cracked it out, as they say in English, but it's like spot on really, and this is why we keep repeating it on the show again and again, and I don't see any harm in repeating it. So you just summarized in a very nice way and I would say in a simple layman terms, as they say, like no one could not say no, I didn't understand what the Andong is saying. So thank you very much about the product market fit. So you're solving a real problem and you explain what this problem, how you're solving it, like what's the outcome? 10x or save time 100%, and the way to build things. So all are these very valuable and great insights, I would say. 

0:36:15 - Yandong
So, andong, final thing where people can find more about you and about Connectly yes, they can go to connectlyai to find more information about the business, and they're very welcome to find me on LinkedIn. Send me a connection request if they want to get in touch, or my email yandongai D-O-N-G at connectlyai. 

0:36:44 - Mehmet
Great. I will make sure I will put this information in the show notes. Again, thank you very much, andong, for the time today and for this great conversation. I really enjoyed it. I learned a lot also about how AI can benefit us in the retail and e-commerce business, and this is something what I love because it's it's a like close to my heart when we talk about also small businesses. So it's good to see that we can leverage AI not only for the big things, also on the small scale as well. So thank you very much for sharing this and I advise everyone to go and check on connectlyai. 

The URL will be in the show note and, as usual, this is how I end my episode. So, for the people who are here the first time, if you're listening on your favorite podcasting app, don't forget to subscribe, and if you're watching this, also, don't forget to subscribe to the YouTube channel. And if you are one of the loyal followers, thank you again for all your comments, all your feedback, all your suggestions. I read them all and thank you for continuing supporting the show. And again, I repeat this at the end of every episode, I know, but if you are interested to be on the show, if you have that great idea, you're doing something really different and you want to share it with the rest of the world. This is a very open place where we can have a casual chat, as we did with the unknown today, so don't hesitate to reach out to me and we can arrange this. Thank you very much for tuning in and we'll meet very soon. Thank you, bye-bye.