222. “Leveraging AI- Not a Google Search” with Ricky Ho

Dawn talks with Ricky Ho, founder of San Francisco AI startup SourceReady, which helps retailers, brands, and importers automate product sourcing from idea to factory by identifying suppliers globally (including some in the US), calculating tariffs, and even negotiating with factories using AI trained on millions of factory records, supplier disclosures, shipment records, and customs data. Ricky explains that most small businesses misuse AI like Google search, while the real power is in reasoning and execution through persistent AI “agents” with memory that can work toward goals using tools. They discuss trust and verification, emphasizing AI shouldn’t make high-stakes decisions yet, but can do substantial work with human oversight like managing an employee. They also cover AI’s growing role in customer service, recruiting, accounting, legal, sales/marketing, and retail, and suggest staying current by asking AI for updates and sources.

Who is Ricky Ho, and whom does he serve? (00:52)

Sourcing Pain Points (01:44)

Domestic vs Overseas (02:26)

AI Not Google Search (02:57)

Chatbots vs Agents (04:59)

Trust and Oversight (07:53)

Managing AI Skills (11:14)

Customer Service AI (14:26)

AI Sourcing Workflow (20:48)

Industries Up Next (24:34)

Vibe Coding Boom (26:34)

Adoption Barriers (29:16)

Stay Updated Using AI (30:43)

Ricky Ho’s Information:

Website:https://www.sourceready.com/

LinkedIn: https://www.linkedin.com/in/ricky-ho-a25a05104/

Transcript

Intro/Outro: 0:03

Welcome to Business Mastery with Dawn Kennedy, your quick under forty five minute dose of expert insights and strategies to make a positive impact on your business and life. Let's get started.

Dawn Kennedy: 0:15

Hello and welcome to this episode of the Business Mastery Podcast. So AI is everywhere. AI stories are dominating the news, and lots and lots of small businesses are using AI, but they're using them as a Google search. So I have Ricky on here today. We're going to be talking about and why we're using it wrong, adjustments in how we use AI to So thank you so much for joining

Ricky Ho: 0:45

Of course. Thank you, Don, for inviting me.

Dawn Kennedy: 0:48

Can you tell everyone who you

Ricky Ho: 0:51

Yeah. My name is Ricky and I'm the founder of a AI startup in San Francisco called Source Ready. We're a venture backed company focused on helping large retailers, brands, importers automate their product sourcing process. So you can think about this as factory fully automated. So everything from helping you to helping you find the right help you calculate the tariffs. And then ultimately, even the AI factories for you. That's ultimately what our tool I've been working broadly in the space for over ten years now. I started a company before this, also in the supply chain technology space, which I sold two and a half years ago to a large private equity called Tree Lab. And then after selling it, I I guess the main pain point that many small businesses, even and challenge finding the right And right now, with what's happening globally with tariffs and everything, it's even making the whole situation more complicated, right? So if you want to make, let's want to go to China for that? Or do you want to go to Mexico? Or do I go to Vietnam? And all these different benefits and detriments. And even if, let's say you did figure out that you want to go to Vietnam, how do you actually find the suppliers contact information? How do you know which suppliers And I think this is something that we believe AI has the huge potential to solve and can really make this entire journey much easier.

Dawn Kennedy: 2:21

Amazing. So do you also do domestic So I just want to make sure So we know that there's a lot of complication with language barriers and contract terms and things. If we were to go international, but those same barriers could be here if you're crossing state lines.

Ricky Ho: 2:37

Yes. So we do have suppliers I do think that the challenge with sourcing in the US is that there are only certain categories of products in which the US still has a manufacturing base. So yeah. But yes, to answer your in the US.

Dawn Kennedy: 2:52

All right. So we titled this or we talked So let's just start there. Because people are treating We'll just call them general I don't know who everybody uses, Because there's so many of them now, just general chat bot and they're using them like a Google search. So can we just kind of talk the search and an AI prompt?

Ricky Ho: 3:18

Yeah. I think the number one difference is if you think about how Google search works, is it tries to surface the best resources or links for that particular query. And with AI, you can chat with AI and kind of use it for a similar function. But I think AI is a lot more powerful than just surfacing information. So I think the biggest reason hype around it is its ability to to execute certain actions. So I think most businesses just because there when they see that that's all the AI can do. But I think the companies or individuals that are using AI, the best is they actually think of AI more like a digital employee or like a digital assistant. And if you think about an assistant, searching for things is just one of the many things that an assistant can do for you. And I think if you broaden the actually, other than surfacing example, integrate with the take the next step. A simple example of this is you find me suppliers in Vietnam That's one thing that I can do. But if you want to take a step further is why don't you actually have AI, reach out to those suppliers and negotiate for you? And I think most people don't even do that. And that's part of the reason why the Gemini and ChatGPT and these tools, they're great, but they're actually a myriad of other tools out there that are AI enabled that can do even more than the general chatbots, for example. This is why we exist, right? We want to enhance the a simple search. So I think that's one thing. Second thing is there's AI that So chatbots are typically the people interact with. But I think the drive since of where the industry is going So if you think about the difference between agents and chatbots is chatbots are not persistent. And that's more technical term. But essentially what that means is any time you want the AI to do something, you have to prompt it. You have to explain what you But if you think about if you Imagine if your assistant forgets everything that you told it about every single time you ask it a new question, and you have to re-explain yourselves every time. That would be insane, right? So agents actually, first of And second of all, you can twenty four over seven, but in of work that you predefine. So you can give it a high level Say, I want to figure out how to find ten customer leads in the next six months, and these are the places in which I typically do so. One way to do it is just to ask That's just like the normal, like the traditional way people do it. The other way is I'm going to give my agent access to all these tools. Google search might be one of use other tools. And I'm just going to let the AI how to do it right. Even maybe I don't even know how the AI doesn't know how it's It's going to test different things until it's able to reach its goal. So where we see the industry telling the AI what to do, you give it the tools to do it, and on its own and execute just like And I think that is really where AI has has a really direct impact on the business, because then you start treating it less like just a glorified search engine. You start treating it like actually an end to end automation platform.

Dawn Kennedy: 6:54

There was a recent story, even on the consumer side, where a programmer built an agent for himself. I don't know what. Excuse me, what he did exactly, on the news. It actually made the news that a car and this thing went out, dealerships and negotiated like dollars off the sales price. Obviously, it can't enter into imagine being able to do that the scenes you wake up or you morning it's like, oh, I have And the more variables you feed and the more that you're able to niche down. And what I have found also research quite a bit, is that even know that I needed to So let me ask you, as far as the I can tell you from somebody who business and has to source There's a trust factor. There's a trust factor that this supposed to do. Because unfortunately, the news, It's only telling you about when it goes wrong, uh, or when it goes and does things it's not supposed to. I rarely get to hear these purchase story in the news. So can we talk a little bit about trusting these agents because we know that they can come back and they can go, oops, I'm sorry. Or you can say, oh, I misunderstood your question and it'll tell you that, but it's not a sentiment being in the fact that it'll recognize its mistakes. And I think that's where there allowing this back and incredibly efficient and likely the processing time and things, to protect margins.

Ricky Ho: 8:52

So I think you brought up a very I think I have two things that I think, from my perspective, would the way to think about this, number one is I don't think AI right now is mature enough or consistent enough to actually make high stake decisions. So as you mentioned that car example, the AI is not actually sending out the purchase order or actually purchasing the car, right? All it's doing is it's getting you all the offers in a way that makes it easy for you to make a decision. So right now, I think in terms of the tasks in which I would give AI free rein to do our tasks in which no decision is being made. So that's number one. Number two is humans make I think I think people are it's a new form of entity. And it's almost as if it's like a how like looking down on AI is somehow going to save careers and stuff. Like I, I completely disagree. Like I, I think that humans will value because of the fact that So whether it does something AI doesn't get anything out of It's basically servicing some So I think if you think about it were to hire someone, they could How do you guarantee? There's no guarantee for That's why you check their work, Even if you were to hire a human check their work? Are you just going to believe So I think if you put it that way, then it really comes about like, how do you become a good manager? So like being using AI similar have to know what tasks you And you also have to understand And checking work just means how do I get to the conclusion it got to where what steps and traces did it do in the middle for it to come to that final conclusion? And you have to check that to logic you have. But I think it's you will miss messes up zero point five of that zero point five percent, ninety nine point five. I think that doesn't make sense It's just like saying, I'm going messes up once, I'm going to good things that he has done. That doesn't make sense. Like you have to look at it from And I do think that's the way these agents or AI.

Dawn Kennedy: 11:09

Okay, so now we're going to get its work, right? So when you said earlier, AI going to approach the task or it human, is there some resistors? Are you seeing something, especially with more seasoned business owners who have been doing this a particular way for a while, where they're like, I don't maybe feel comfortable managing it? Or if I can't follow the logic the future that there will be management sort of function across several companies? I could kind of see that sort of that are niche into particular And I can tell you the idea of managing AI could feel a little overwhelming if you're not somebody who's to protect savvy or feels comfortable with the platform. So what do we do around our own being able to manage an agent?

Ricky Ho: 12:11

Yeah, I think that will become of the twenty first century. It's just like saying today, if computer, what will happen? If you don't know how to use a Then it's probably going to be in a desk environment. I think AI is similar right now. We are still in the very early So as a result, if you don't could probably get away with it fully on board yet. But if you close your eyes and wait ten years, I do believe that every company in the world is going to be embracing AI to the full extent that we embrace computers today. And this is why this is a skill You just have to learn it. If you don't learn it, then takes to be successful in this So that's my first perspective. And I think the second thing is easier to manage in humans. Because as I mentioned, when you things outside of just objective There are their emotions, right? Every human has different You have to take that into And I think the benefit of AI is you don't have to think about that. The AI is only wish and goal is to serve you for whatever goal you have. You don't have to think about You don't have to think about And I think like when you talked work, AI actually does a better what they did. So if you use any AI agent, it the way, the reasoning trace, Oh, it didn't work out. Okay, now let me try this. It actually speaks out loud what it's doing throughout the process. And I would say it's not that hard to understand because it's just speaking natural language, right? It's literally it's explaining step of the way and how it got I think with humans, and going most people actually don't do a explaining what they did. They'll give you a conclusion and you have to go ask them proactively, how did you get to that? And this is why if you look at AI agents on the market today, almost all of them show you reasoning traces. They will show you exactly their And I think this is something how to interpret. But I don't think it's like technical that only people that

Dawn Kennedy: 14:20

Okay, perfect. So let's get to the next place listening to this episode and So there are particular things that AI has had some backlash on, like the customer service do loop with several companies like Verizon or whatever, and people are like, hey, I don't want to be stuck in an AI do loop in AI chat, right? What you are proposing, what your company does is not consumer facing, not customer facing. Where do you see or do you see to be customer facing? Or do you really see that the benefit is going to be this behind the scenes, the non consumer facing or non sales facing roles? Because if you can reduce your back end cost, doesn't that mean you can hire more human facing people.

Ricky Ho: 15:10

Yeah. I think both are going to There's going to be AI for back And there's going to be AI for kind of consumer facing roles as well. So AI for back office is obviously the most immediate ROI, because it's just so obvious that if you can automate something that traditionally a human would do and for one tenth the cost and have it work like a superhuman in the background, like, of course that makes sense. Consumer facing is a little stakes are higher. You don't want AI to go off the piss off your customers. But having said that, JetBlue the largest consumer facing already fully in their business. If you were to call Chase's AI agent first before you talk Same for JetBlue. They use AI tools for the entire They only send you to a human if the AI is not able to answer the question. So there is a metric that people It's called, um, sorry. What's this? Basically the term I forgot the technical term, but it's like it's what percentage of questions can the AI answer on the first run without human intervention. And and that rate has massively So like traditionally, like for like two years ago, most of the time people would just transfer to human. They'd be like, I don't want to Just get me a human. Right? But now what you can start and so natural that sometimes talking to an AI. The, the voice and the intuition questions is very human like. And I think this is also where Like a company called Sierra or I don't know if you've heard of that company. They.

Dawn Kennedy: 16:52

I have not, but.

Ricky Ho: 16:53

They are worth thirty five Wow. And yeah, they build AI customer used by some of the largest in the world.

Dawn Kennedy: 17:05

And are they doing that to In other words, if you're calling to see when we're open and it's not obvious from our website or our Facebook group, I have my own thoughts on people who ask those kinds of questions of customer service people anyway. But if it's something related to require a specific set of facts, be more successful. Whereas if it's a fact specific maybe my ticket number or my have you, do you see that as the To humans, because this is the As humans, we like to always think that our problems are unique, right? We always want to say, no, you Yeah, that's such a human thing. To think that because we have an issue, it's an issue that only we have. And so I'm kind of curious, from to see AI being also able to heavy specific individual issues general issues that maybe a

Ricky Ho: 18:19

Yeah, that's a very good So the example you just brought up of checking the ticket number and then maybe servicing a refund and seeing if the specific request actually adheres to the airline's refund policy. So let's say a storm happened and some for whatever reason, you want to refund your ticket and you call the customer service and say that Sierra and these AI chatbots today are already good enough to actually say, hey, sir, wait for a moment. Let me go. Let me check your ticket number. So then the AI can check the who you are, what? When did you book and all this And then it will ask you some And the way it asks you the would, it would first check if matches the airline's policy. So this AI has already been trained to understand what is the specific policy it needs to adhere to. So I think to answer your question, yes, one hundred percent. Obviously there are going to be like extreme edge cases where the AI will not be able to answer. Let's say the human gets super pissed and starts yelling at the AI. I think at some point.

Dawn Kennedy: 19:22

That doesn't happen, does it? Is that I don't know.

Ricky Ho: 19:25

Like still presented if that AI is at the level of calming it's like a very action oriented example, essentially you need certain facts or certain data to some form of policy that has asking additional questions if So I think AI one hundred percent can do that today, and I think it could probably even do it better than some humans simply because AI has a better memory recall. You could imagine an airline could have hundreds of policies, maybe other. And as a customer service human, They have to go check and Oh, is it okay for me to give But AI can answer that very information preserved to it.

Dawn Kennedy: 20:17

Okay. So really what we're saying is the normal regular run of the mill transactions that a lot of us do, AI is already doing it within the parameters of policy. So it almost becomes this idea where humans really only need to get involved, or at least in the future, to develop if there's an exception to policy, if there's just something really weird or something that the AI can't recognize.

Ricky Ho: 20:41

Yeah.

Dawn Kennedy: 20:43

Awesome. So let's go back to sourcing Right now, the process probably for a lot of companies is they have to maybe use the chat AI and they say, find me a supplier. And when they find the supplier, And then they either have to they have to send emails to all So really, what is the I'm looking to do this? This is and I'm sure that you need to put in, right? You need to define all the point you're looking for. And then this can even send inquiries to potential suppliers.

Ricky Ho: 21:25

Correct. So I think there's could save a lot of time on. One is search. And it's the way it does. The searching is not like a Google search or like even asking. ChatGPT is first of all, we've trained our system on millions of factory data that we get from supplier disclosures and shipment records. You actually can't find this information on Google, because a lot of these factories barely have publicly available information. Like they don't have a website. Or even if they do, it's very bare bones and it's very hard to understand if they're actually reputable. So we actually get data from world to tell you what customers Let's say you wanted to find a for yoga pants. You could literally search that And our AI can tell you exactly who are the suppliers that work with Lululemon. And second of all, it's able to go through millions of suppliers and rank all of the top suppliers that match your specific requirement. So you don't have to manually they're a good fit. And then the third thing is and the outreach to factories. I think this is probably the most time consuming part of sourcing a product, because every requirement is different, and you basically have to negotiate back and forth with multiple suppliers. And there's a lot of waiting and And essentially what you can have AI do is you give AI a goal. You say, hey, I want to source this product for five hundred pieces. I need a target price of six Go out there and figure out what suppliers are the best match for this. And the AI will first build a list of maybe fifty suppliers that best match this high level requirement. And then it will do the heavy there and going back and forth it has enough information for you want to work with. And from what we've seen, what pool of suppliers that you have because if you do this manually, to reach out to fifty suppliers. Most people just reach out to But by reaching out to more, you basically increase the chances of you finding an even better supplier with a better price or better terms.

Dawn Kennedy: 23:37

Yeah. And it's important that and those types of things, And on the bottom of every AI that I'm aware of right now, there's a little teeny tiny disclaimer and it says, AI can make mistakes.

Ricky Ho: 23:50

Yeah.

Dawn Kennedy: 23:51

Please check your work. Yeah. Because to your point, as you said, software cannot be responsible legally. It cannot be held responsible if and it takes a business, the ultimately responsible, right? Yeah. Because the software can't be held responsible legally or otherwise. So the idea that AI, because of second and third and fourth that nobody else can get to, to It's a, it's a trust but verify So you're able to do that. Yeah. Which industries other than like supply chain sourcing and airlines, AI customer service, where else do you see industries needing to get trained now to expect this within about the next eighteen to twenty four months. The supply chain products. But there's other industries uh, where it's not coming here. Right. And yeah, based on our

Ricky Ho: 24:55

Recruiting for one hundred I think accounting accounting is lot of increase in adoption. Legal is, I think, the last could be helpful with. And now we actually see a huge companies helping lawyers save a Let me think marketing. I think sales and marketing is I think the sales job in the focused on nurturing research or mass outreach. I think that will be jobs primarily done by AI and I guess even retail. I don't know if you've seen this, but there's a lot more AI enabled hardware where physical Stores. AI can help provide better recommendations of products to customers and potentially even have in the future, robots at retail to help out with some of the repetitive customer, let's say, returns or issues related to that. Yeah, I think AI is going to go don't think it's gonna have the think it does. I actually think human will do will replace a lot of the humans don't like to do anyways.

Dawn Kennedy: 26:09

And that to your point, and I'm there's been a lot of news and programmers have been laid companies because of AI, but I that a lot of these programmers understand the technology are independently, niche companies. There was an AI agent. I saw it on the news Where one AI agent was like a personal assistant and was booking travel. I think you might have maybe you the travel person or travel switched to another language to make it faster. And what probably would have taken a forty five minute phone call was literally done in about six minutes. All the details, all of the stuff, everything was cut, And it was very interesting to watch because those two independent agents were built by people who worked for other large companies. So I'm wondering what your again, if it bleeds, it leads. They're yelling about all the how consumer computer science But do you see this trend Yeah, these people who are trained on this stuff are going to go out and actually serve in niche markets and build really cool things like yours for sourcing to help humans across industries and across different task needs.

Ricky Ho: 27:40

Yeah, I think one hundred And I think what's exciting AI is good at coding now too, also building incredible tools So now I don't know if you know you can build your own AI understanding how to write code to code up an application. And we're seeing that across So there are certain things where, let's say you have a very tailored system that you need to build that doesn't exist in the market. Traditionally, you would have to it because you're not technical. But now, because AI is so good at coding, you can just tell the AI to build the application for you. Obviously, for more complex doing on sourcing, for example, little more technical expertise But what ultimately is going to there's going to be so much more the barrier of creating software So I think that's a very There's just going to be so many that businesses can use to And I think that's just going to

Dawn Kennedy: 28:51

Amazing. So do you think the biggest having a trust, or is it really and they feel a little unsure Where or do you think it's both? Like, where do you see the biggest eighteen to twenty four month? What do we need to overcome to

Ricky Ho: 29:11

Yeah, I think one is skepticism. I think people are overly aren't willing to really In real business use cases. They use it more for fun or more won't trust the AI to do actual think is actually a mistake. And I think we need to embrace And I think it will have a very positive impact on both the lives of the workers and for the business. And then I think the second that aware of AI's improvement. If you think about how AI was compared to now, AI is improving people it's very hard for And they don't know that AI has So I think my advice would be AI everywhere that you can and And then number two is be updated on what's happening in the, in, in the AI industry, because this is one of the fastest growing and rapidly evolving industries in the world. And every couple months there's There's going to be a new skill do even a few months ago.

Dawn Kennedy: 30:18

So what sources would we want to Because if you listen to the Yeah. And the business news, even CNBC at enterprise level and things shareholder interests and stuff. Where can we get the real news without having to read an abstract from four engineers to keep track of what AI is doing and help us make the decision to maybe trust it more, implement more.

Ricky Ho: 30:46

Yeah, I would use AI. That would be my suggestion. I wouldn't read a lot of these news, they their headlines are I would just ask AI every couple developments on AI? What are new models, new tools And obviously for every business about in AI are different than So you can be very tailored in You can ask AI. Let's say you're in the are there any new AI do X job better? And AI can maybe recommend new models that recently came out.

Dawn Kennedy: 31:23

All right. So we're going to ask the AI to That's interesting. The nice thing about AI, and market research is you can say, Right. Let me check what's going on. And there have been more than back and I had hey, that had a Find me another source or I can't use it for my article or whatever. So yeah, I think to your point, You listed all the industries and can you tell everyone where they can find you and learn more about your AI agent, particularly if they're in the products business and need suppliers?

Ricky Ho: 32:03

Yeah. So our product is free to You can go to source ready. You can find me on LinkedIn. So if you were to search Ricky Ho source ready, you'll find my LinkedIn. Feel free to connect with me. And yeah, I think that's the

Intro/Outro: 32:16

All right.

Dawn Kennedy: 32:17

Perfect. So we'll put that down inside If you are a product based manually finding suppliers, this move to a more efficient. And it sounds like a much deeper well of potential manufacturers and things. Because I do think you're right Yeah, three is good enough, That's kind of what we've been you don't go to fifty because So I think this is an incredible And I thank you so much for coming on here and talking about, hey, AI is just not Google. So thank you so much for being

Ricky Ho: 32:54

Awesome. Thank you so much.

Dawn Kennedy: 32:55

All right. We'll talk to you all next time Business Mastery Podcast. Take care.

Intro/Outro: 33:04

Thank you so much for listening to this episode of the Business Mastery Podcast. If you want to learn more about dot com and you can now check us course, any of your favorite Take care.