Amazon's Customer Service is Fake - Here's How to Beat Them

01:23] angelina: Hi, summer.

Hello.

Hi. Nice to meet you. Yeah. Today I'm very pleasure to learn more about Task, which is a product that you're building, right?

Yes. We're honored to present here. Yeah.

Awesome. Can you introduce yourself to our audience at Toset ai? Who you are, what are you working on? Yes, briefly.

I'm co-founder and CEO at Task Monkey.

Task monkey.ai is a AI price watcher and negotiator. We help you online shopping price drops, and then we negotiate that price drop for money back from the customer service.

That sounds like an amazing product. I have a series [00:02:00] Lightning questions about Task Monkey. So my first question, how old is Task Monkey?

Two months. Two months. And. Half. Yeah.

So basically Test Monkey is a baby right now. Yeah.

Yes it is.

Let's say your grandma calls you and she asks you, what exactly does your company do? How do you explain it in one sentence? Grandma, I'll always find the best

price for things that you want to buy online.

Does your? Sure. That price.

Does your grandma shop up online?

Oh yeah. That's the main question.

That's a great question. Yeah. Okay. Do you remember how many refunds has Task Monkey platform already processed so far?

Uh, right now we fund over $120,000 for our customers, and we secured over 30,000 US dollars.

[00:03:00] That's amazing. What's the craziest refund story that you've heard? Like the biggest amount where someone was probably like, oh, no way this worked.

I think there are a lot of stories that people, the price drop was like maybe a few dollars, but after the negotiation, the AI was able to secure over like much bigger amount than the original price of the product.

For example, we had a customer buying, I think a candy on Amazon, and it was like $10 itself, and the price drop was around $4, was like 40% price drop. And after that negotiation with the customer service, they refunded I think 20 $20 promotional credit. If you think about the, the proportion wise, it is 200% of the original price.

Crazy.

Yeah. And so that's [00:04:00] about the like the proportional,

highest proportional

thing. Yeah.

What about absolute amount?

I think there are a couple deals that we secured over $900, and those were, most of the time it's electronics, whether it's like robo rock to use those like automatic vacuum. They have a lot of promotions, all and off season, and I think the biggest one we secured was a ROT 900 for the A robot.

And there were also, I think a couple deals, a couple refund we secured, which is over $500. Those were like furniture. Yeah. So I would say Amazon sellers were pretty generous.

Yes, as a lot of dynamic pricing as well, the prices fluctuate a lot. I didn't know that. I never looked back. After I purchased something, I never looked back [00:05:00] at pricing again.

So I didn't know about this

exactly. But it's hard for you to time what is the right time to to buy a product,

and you have to monitor the price change all the time as well. Yeah. In order for you to know there is a price change that happened. So can you walk me through this? If I buy something on Amazon today and I install Task Monkey as a Chrome extension, and then what actually, what happens after that?

Right now, after you purchase on Amazon, you start Task Monkey and the monkey will scan your pots to 90 day orders and will find the price drop. Then start to chat with the customer service using the AI tools and negotiate for that money. We'll also generate a summary of how much money we ask back successfully, and whether it is going back to your bank account or it's a promotional credit, it's gift card [00:06:00] balance.

We'll give you a summary. So basically you push that start saving now button and everything's on us, and you'll just be waiting for a report of how much money will go back to your account.

That's amazing. So it's literally clicking on one button and then Task Monkey will scan my pass. You're saying 90 days of transactions, and then look for price drop, and then you will negotiate with Amazon's customer

service

and for refund or promotional credit, whatever that is.

Yeah, exactly. That's what we want. We want an autopilot, we want an AI autopilot that does the job for you and. For the money back for you, instead of like a copilot that you have to watch and monitor all the time.

Yeah. Yeah. So everybody's building ai, customer service bot, and you are building an entire customer service bot.

So you're completely on the consumer side. Right.

[00:07:00] Exactly.

Yeah, that's very interesting idea. I'm not aware of like any, do you have competitors

right now? We have different kinds of competitors, like do not pay.com. I think they're already pretty big. They started around 20 15, 20 16, and they also stand from the customer's perspective.

They dispute your bill, but I think we're very AI native comparing to them. We rely a lot on this AI bot to fight back.

I do want some money back if there's any price drop. Especially for larger items, right? Like the robot cleaners. I do have the one of those. Why not? But never engaged in anything. 'cause it's time consuming, isn't it?

Talking with customer service can be a pain. It's a right.

Yeah. I don't hate tech, but this, in this past 10 years, we have seen so many AI chat bot coming from the e-commerce side, from the retailer side. And these AI bots now make [00:08:00] things a little more too complicated than I think they were designed for now, whatever is your original request, you need to spend around 15 to 20 minutes to initiate a talk with the the customer service.

You need to navigate through all these menus to select whether it is a chat bot or a real human customer service, which department you want to talk to directly. There's a lot of choice before that talk start, and even after that talk starts, you have to make those choices. Sometimes again, or you have to tell the person inquiry.

Yeah, your inquiry multiple times in that process of talking with a customer service. Yeah, that's exactly why we started this thing. 'cause we feel like all these three [00:09:00] retailers are using much more complicated and fancier technology. But on the other hand, customers are left alone. They. They have to deal with these technologies by themself.

So Task Monkey will handle chat with another bot, including handling the escalation to a human being where most of the refunds will actually happen.

Yeah, so right now our AI talk to the other ai and we also talk to the other real person.

What's the average lens of these refund conversations? How are we saving our users?

Like I said, I think somewhere from 15 minutes to 20 minutes, it's an average for like median conversation land. And so there are a couple of our conversations goes to 40 or 50 minutes. It was a lot of waiting, just like in between the conversations. The other customer service say [00:10:00] they need to check with their boss, and then there's 200 seconds waiting time.

Yeah. That happens a lot.

What's your AI agent's conversation Success rate, like refund success rate?

Right now it's over 90%. That is actually pretty conservative. I think we're close to 95%. At this point,

that's pretty good. I'm not confident I can be successful in talking with the customer service, asking for a refund.

If they say no, then I'll probably say, okay. Alright,

good. That's another thing you need to think about first, your time and energy that you put into this, does that really worse? Is that primary job really worth your time and energy? That's the first question you will debate. And the second question is, people really don't want to get into these negotiation thing for the little money most of the time.

Yeah. But accumulatively, they're quite big.

Yeah. Yeah, [00:11:00] exactly. I actually just run it yesterday and I got 10 bucks back.

That's nice. Yeah,

why not? I'm surprised like, you know, wow, this, this actually works and I was watching the whole conversation going through. It didn't interfere anything. Yeah. So it works.

Yeah, and for many retailers, they actually offer this policy. You didn't know that Costco offer 30 day price match. Home Depot offers 15 days, target offers 15 days, and that some of these retailers, they actually offer 10% off. If you find a price match.

Mm.

Yeah. It just, people don't have time to figure out these policies and they don't have time to submit the claim.

Right.

Perfect. For AI to do.

Yeah, exactly. Yeah. Finally, an AI on the consumer side. Oh. Because the e-commerce was the chat bots. They're more defensive, right? They don't wanna let you talk with the real humans and deal with complex situations. [00:12:00] So now we don't have to do it. AI will do it for us. Did it ever happen that the other side of the customer service, the human recognizes that they're talking to a bot?

So it's interesting that we recognized they are bought. Okay. So on Amazon customer service, they pretend to be a human being and they each own a real name. But after tens of thousands of conversations, we realized their patterns and those patterns doesn't sound like real human being. And we discovered that they're also chat bots.

So consumers wake up. Think about the time that you wasted talking to all these chat

bots, right? Yeah. And not then get nothing done.

Yeah. And they also prolong that waiting time to pretend they're human beings.

That's me.[00:13:00]

Yeah. But it's now, it's interesting watching our AI to fight with their ai. That's fair.

Totally agree. Is there any fun conversations that you've seen in your history that the AI is pushing forward for the refund? 'cause as a human being, like you said, it's $1, $2, I'm probably not gonna push for it, but based on what you just shared, the AI will actually be able to push on your behalf.

Yes. I think our AI is actually very stable emotionally. They don't get distracted or disappointed if the other side doesn't offer anything or doesn't offer as you expected.

Right.

The AI would be pretty. Persistent and consistent in their tones saying that I see this price drop and I do want us this back, and we see the other side.

Sometimes we're reluctant and they're saying, we don't have this policy

and.

We see there were conversation [00:14:00] that the other side, the customer service saying, okay, yeah, you can return that product. That's the only way for you to secure that price drop. And then our AI was saying, okay, I am going to give you this math about how much does it worth for me to send it back.

And then you, you guys handle that shipment and you have to handle all these operations of. Things returning to the inventory and finding a another product, which is the same as the one that I'm returning and we spend even more on the shipment. So that AI is basically just making the other side, not really confused, but just really getting annoyed at that.

And at the end there were a like a. Prolonged waiting on the other side and they're saying, okay, for this one time we're offering this promotional credit back to you so that we can both save some hassle. [00:15:00] And our AI were, yeah, we're like very glad to accept that and, and that conversation right away I think is actually pretty smart.

Yeah. Just watching that conversation carrying on, see that AI is like really good at these things. They don't really care how much money they're asking back, so they're never really awkward for requesting price drop. And they are very distant, like we said. And there's another example. The story is our AI requested the price drop and the other customer service said, okay, it's my birthday today, so I'm giving you a little more and I'm happy to talk to a customer like you.

And then our AI was like really sweet. And turn to say, oh, happy birthday. I feel so lucky to talk to a customer [00:16:00] service who is having birthday today. And I feel like you bring me luck. And that makes the customer service happier,

being happy.

Yeah, and that's, that's what we pictured at first. And apparently that conversation was shorter because they offered money back right away.

But after thousands of conversations, we discovered that pattern. They are different Amazon customer service saying the same exact sentence, it's my birthday today. I'm offering you a little more.

I'm surprised.

And then you realize that, yeah, Amazon put a lot of effort pretending that their are real, real human being.

Oh my goodness. I was just about to say that. Does that mean that they're more likely to be human? Looks like, no, it's not true. Or maybe it's in their employee handbook. If it's your birthday, you gotta tell your [00:17:00] customer that it's your birthday. Could we

not?

Yeah.

Yes. We went back and saw that those words are exactly the same.

Yeah. I suppose those are ai. I think.

Interesting. Yeah. Surprising. But it's okay because Task Monkey, our AI is also shameless. So shameless AI to the other shameless ai. Yeah.

I'm thinking that's like, um, definite trend that each consumer would have their own AI team to deal with the retailers or like any other party AI team.

That's the future.

It's hard to imagine how that works. I definitely can imagine a world that as a consumer I can have some AI agent to help support my daily life and live a more efficient and beneficial life.

Right. That would smooth the whole process out so nobody would feel shameless [00:18:00] where they would never feel awkward.

Yeah, exactly. No, no issue at all. Go ahead. I, I saw some of the transactions I had. There's 67 cents price drop. Task Monkey still did the negotiation for me for that 67 cents.

Yeah.

Why not? I wouldn't do that, but go ahead. Why not?

Yeah, but think about the price drop when there's a couple hundred price drop.

And that's a little over 30 days that happens to our customer. It just passed that 30 day return window, but the price dropped a lot. And sometimes customers get angry, but the other customer service, like the one that we found out or pretty much supported by ai, they don't care. That's an even more awkward conversation.

You're like, you're irritated by. And AI customer service, they don't respond to you. But now if you let our AI deal with them, you can actually initiate that talk multiple [00:19:00] times. You can try at different time, and you can just let that AI to deal with the other ai. They would be, like we said, emotionally stable and persistent on requesting your consumer interest.

Is there any strategy behind the scene in terms of personalities that you give your AI agents when they're doing these refund negotiation on behalf of the clients? Sounds like it's neutral, but you also mentioned being angry could help sometimes.

Yeah. But right now we're act tweaking on that. We want to give our consumers this option to choose your tone.

'cause sometimes our customer want to, sounds like friendly. And some of our customers just doesn't care what tone they're using. 'cause at the end of the day, this AI is representing the customer. And so sometimes our customer want us to [00:20:00] behave in a certain way and we want to give our customers a little more choices and flexibility.

In terms of that to actually, a lot of factors would affect the success rate that include this to, and also include like the timing of the request you sent, especially if there's a price drop right now or there's a price drop coming in. Just like right now the prime game is coming in, so the prize actually would be affected and.

There's also maybe you have requested multiple refund in the past week. The frequency, a lot of things would actually affect the success rate, and I think that's. Knowhow that we figured out through tens of thousands of conversations.

And would you give that insight to the customer? As a [00:21:00] customer? I don't know.

When should I initiate such conversations? Because I still have to click on that button to have task Monon review all my past transactions in the last 90 days. Right. So it will be nice if you can tell me, actually, this is the time to negotiate, you know?

I see what you mean. I think right now, so try it when there is a big promotion going on.

That's first. So that all the sellers actually couple

days.

Yeah. The sellers are, they expect to spend this marketing money and second is just be persistent, but emotionally stable. Because, yeah, they're different outcomes and you don't want to sound like you're. Angry but so that you expect a lot and you are like forcing the other one to go into this deal with you.

Emotionally stable, state [00:22:00] your request very clearly at first, and I think that's the best a consumer can do.

So Task Monkey can basically handle what you mentioned, handling the conversation. Be neutral, be persistent, be shameless. Then as a consumer, I can choose the timing. I can choose like day of time, day of week, or the holiday to initiate a conversation.

Yeah,

yeah. Sounds great. Yeah, these are a little tips that's actually quite useful for for me to consider this. So I wanna ask you, let's go back to your founder journey. Do you have an eureka moment about this specific product? Did you have some personal experience that motivated you to build this?

Yes.

Before teston, we built several different AI agents. We were always trying to solve our own problem first because we at first, uh, noticed how powerful AI is. It could [00:23:00] easily customize a lot of different things and bring that. Cost of development and customization done a lot. And so I think what's different from us, the other founders is that we think this is a time to build something for consumer because the cost to customization is a lot cheaper.

And I think consumer wants are like they have long been waiting for technology to bring this balance back. By which I mean, so like we discussed. Previously there were AI chat bots emerged in the past 10 years. There were dynamic pricing and there were a lot of technologies to understand consumers.

Either in the offline space or in the on the online space, but there were not that much knowledge or education for consumers about what's going on. And so you were like forced into the situation [00:24:00] of having to accept or what was offered. Like you don't know how much inflation has affected your price, right.

You don't know how these dynamic pricing are trying to harvest all the revenue below the supply and demand curve, and you were learned and studied both offline and online. At some point, your customer profile would be matched and so you know, consumers are

helpless. I think we're helpless.

Helpless,

right?

Yeah. There's nothing that's out there that's protecting us in a sense. That we are aware of these big data studies or researchers or algorithms. My personal experience, if I use MacBook or I use like a different computer like my mom's computer, which is not a MacBook, the pricing is different. Like I, I'm like, oh, right, like if you use different device, there's different pricing.

So

highest discrimination.

Yeah. Yeah. Total discrimination. So it, it's really hard for consumers to be aware of [00:25:00] those things and. Not to mention to find ways to protect themselves.

Yes. But I think time has come because AI is here. Yeah. Even if it's not Task Monkey today, I think because the development of such technology is getting cheaper and more convenient, consumers are getting smarter.

I mean, this trend is inevitable. Yeah. So should be prepared.

Yeah. Yeah. I totally agree. For your personal journey. I did my LinkedIn due diligence. So it looks like you were working on the street, you were working at Wall Street, and then you're now helping people saving money on Amazon. So my question is, do your parents, do they ask you like, summer, what are you doing?

No.

How? How did that conversation go? They didn't ask you on No question. No question on your choice on career. Tell me more. [00:26:00]

Right? Yes. So my dad is a math professor. My mom's a doctor and lawyer. I think they, they do both have a pretty good career. And so I think, I love to math, thanks to my dad, I love to read all those like Ms.

Madison stories and, and that's how, uh. Went to study math in college and then graduated, went to Wall Street, worked at banks, dealing with all these numbers. But I think at some point you start to think about the meaning of your work, essentially building something that would improve people's efficiency, people's life.

That gives me much more meaning, or that makes my work much more meaningful. Speaking to customers definitely give me much more incentive to work. [00:27:00] That's not saying the finance work is not meaningful. I think people see like value in different ways and from different perspectives. Finance is meaningful, but it's uh,

it's, that's obvious, right?

Yeah. It's not, that's

not miserable. Like right now you're doing something. It's much simpler math. But it's very tangible. 'cause people, customer can tell you like, Hey, I got 20 bucks back, which I only spend $10 now. What kind of satisfaction is that? It's different. And then when like tens of thousands, those customers, then you feel like they're.

Reachable, right? You can reach them versus your customers in finance and in, I don't know what you do option pricing. If you're doing like, oh, I have no idea what people are doing with my model.

Yes.

Yeah.

I think these, the technology world's fascinating because I think they almost all starts with a very simple pain point or like [00:28:00] a, a question, think about Google.

They started because people like to ask, right? They want, yeah, they want search, right? Yeah. So I think a being technology is, is a good way to scale solution and Yeah, and it's so tangible that you can quick your solution on the way and grow with your customer.

Yeah.

Yeah. I think that makes the whole process very transparent.

And yeah.

To you, I feel the same way. You are rewarded by that. Yeah. Task monkey is so young. So I'm curious, were there any like dark moments in your journey? Like about, oh, this is not gonna work. Did you ever have any moment that you felt like something's not gonna work, or something is more challenging than you thought?

It's interesting that, like I said, we built several different products all in consumer world.

How many products did you build before this one? [00:29:00]

I think before.

Okay, when did you start?

So it's been,

yeah, so it's a year.

They're bigger problems and they're smaller problems. They're tangible problems and they're quantifiable problem and non quantifiable problem.

And when we first started, we just found a our own problem to solve, and that's non quantifiable. And you discovered that when you start to put a price tag on your product Yeah. And you realize that there's a big variance in. People's understanding of the value of that product. And that's, that's still a, a valuable product, but it just hard to scale because it's complicated.

Yeah. But I think for Task Monkey, it's quantifiable because the money that AI ask back for you is real money. That's a number. And you can see that AI created that much value for you in that much time. And it's also a very simple solution. It's [00:30:00] autopilot. It doesn't require you Amazon lock in, it doesn't require

any of your

additional actions.

Ideally, it just a very simple step. So that solution is scalable? Yeah, I think so. In our past projects. So right now, the good thing is that it's very easy for you to build a solution, but it's harder for you to find a real market and also to market that solution. And it's also, uh, harder process to put a price tag on it.

Just pricing AI solution is. It's a challenge and I think it's a must of, yeah. The market right now,

selling the product is the harder part, right? And unlike building products, maybe more than three years ago, both technology is still a big component of building a product. Today may be more emphasized, is gonna go to marketing and selling.

Exactly. [00:31:00]

Yeah. And I think another challenge is also understanding the bigger frame. So I think, yeah, everyone is talking about large language models and all these vest technology. As an entrepreneur or as like a smaller team, whereas your standing point or as your position in this AI revolution, I think yeah, that requires a lot of study and understanding framework.

I think in this year I've already seen many companies or many products that were overrided by the evolution of large language model.

Yeah.

And Yeah, but I, I think it's actually quite challenging for the entrepreneurs to understand where these large language models are going and where is their standing point, where is their position in this market?

What we understand at this point is find a vertical, find like a [00:32:00] smaller, like a pain point. You can dig down a little further and you can just focus in that vertical. And so right now in our back end, we're testing this price match and negotiating. AI for Home Depot, for Costco, and for many other retailers.

And that's where we realized that this solution is scalable, both horizontally and vertically. It could be scaled to include multiple other retailers. It could be scaled to include other countries, and so that's a very flexible market, but it's still like a vertical. Pain point.

Where do you see this in the next five years?

I think everything is moving so fast,

right? Agree.

Yes. That yes, that I can see that in one [00:33:00] year we're scaling the solutions to multiple retailers. It may be two to three years. We're, I think deploying that to different markets, but I think in five years maybe we're going, we're expanding towards, are there part of this user shopping journey instead of just like post-purchase.

Maybe we'll be like going a little further.

I'm looking forward. You're building a whole team, like protecting my consumer rights.

Exactly. Yeah.

Maybe, yeah, maybe in a year or so, or six months whenever you hit your next milestone, we can continue this conversation again. You can come again and share your journey with uh, yeah.

So let's say if someone made you CEO of Amazon tomorrow, would you shut down Task Monkey or somehow make it a feature?

Oh, I would make it a feature, definitely. [00:34:00] We, we've thought about that over and over again. Are we really like hitting someone's bottom line, hitting these retailers? But we have seen in this, these like Chinese e-commerce, like Alibaba, they're offering this service for their VIP customers by automatically doing this price match on their own.

And so what we've seen on our platform is that we increased. Customer loyalty for these brands. So think about that as a customer today, whenever you buy on Amazon, you can always lock in the lowest price in 90 days. Would you? Yeah. Would you ever think about going to some other platforms? No.

No. Yeah.

Right. Yeah. But maybe you'll do cross, cross platform. I don't know if you can make cross platform happen. That's even better. It doesn't matter anymore.

We can do cross platform. And we also found out that our customer offered their price match receipt. You know, [00:35:00] that's like a communal effort that make the things even more interesting.

So one of our customer offered their receipt of buying a robo rock online, and he got a very good deal. And so for many retailers, they just need a receipt to give you that offer. Yeah. So our community actually can benefit from that.

I wanna ask one or two more questions, if you don't mind. How do you charge the consumer?

Like, what do I pay you

right now? Or taking a 10% success fee?

Mm-hmm.

So yeah, if we, okay.

You're not charging me monthly. Awesome. Why not? That's like a no brainer. 'cause these are money I'm leaving on the table anyway.

Yeah. Right now we wanna focus on result based. Yeah. So if we get something back for you, we take a cut.

If not, we're not taking anything.

Last question. Everybody's talking about AI replacing jobs. So I wanna ask you a question in this regard. You use AI to build a [00:36:00] business. So do you have any advice for average person who may be average user of AI or even scared of ai? Do you have any suggestions?

Try to build something.

Are you an engineer? You're not an engineer,

but I do code. Yeah, and I think first of all, try to build something in this process. You will discover a lot of useful tools and it's easier for you to understand your position and how these like large language model would affect the process of development.

And yeah, that's step one. Step two is, I think now questions actually go back to operation and marketing and understanding of this one.

So the first, first build something second, try to sell it, right?

Yeah. Just understand how easy it is to build a product. So there's like. Yeah.

Do I have to know basic [00:37:00] programming to build something?

I think you need to have like really basic modularized understanding of coding. And actually there's also a bunch of tools that requires no coding experience at all. No code. Yeah. Platforms are everywhere right now.

Yeah. Vibe coding kind of thing, right? You can do vibe coding your way to put something maybe.

Yeah. Start by building a Hello World app on your phone. And once you see that up and running, you, yeah, you notice that you can build a little more complicated, little more fancier product and just step by step you can sue and. Notice that there's a lot of things that you can build on your own right now.

Then the second question is going back to understanding the market, where is the biggest [00:38:00] opportunity and where do you really want to build a solution that's scalable? Yes,

and this advice is, is applicable to all age group. You're seeing all age group could try it out.

I'm seeing my mom using AI a lot, but I haven't seen her building any AI app.

But yeah, maybe I think I, I can see that in the foreseeable future she's gonna dictate to build an app. I think that's, yeah, that's so you can do everything, but where do you want to spend your time and. Resources.

Thank you, summer. Really nice having you here today. We had a great conversation. We learned about your product.

It's very interesting. It's very different from a very crowded customer service bot space and Right, you're doing basically the same thing, but on the other side, it's not a past that a lot of people have walked. So hopefully we can see more of your progress [00:39:00] and then you can share more on our channel as well.

Thank you.