Felipe Chaves, CEO Of Triplo Speaks About The Benefits Of Ai | The Biz To Biz Podcast

Welcome to the Biz to Biz Podcast, your number one source for all business news. Hey everybody, welcome to this episode of B2B Vault where we're going to dive into AI with Felipe Chavez, who's from Triplo AI, and he's in, um, Rio, Brazil, and he has a new AI software product newish. And we're going to dive in and talk a little bit about AI today because AI is a super popular subject on Google. Everybody's searching around for AI. People know about, you know, OpenAI, ChatGPT, you know, whatever's on Edge, their product Claude, all these different products that are out there. So, you know, AI is everywhere, and we're going to dive in and talk about it today. So Felipe, give us a little intro.

All right, thank you. Thank you for having me, Allen. I'm an enthusiast of tech. My story is like I was a marketeer by design. So I worked for most of large industries that you or your audience for sure know. I worked for Coca-Cola, L'Oréal, and GSK, and so on and so forth. And now I'm developing triple AI. Which is a productivity assistant. This is the way we call it. And I also work for a big data company here in South America. So my pleasure talking about AI. It's a very good subject to discuss.

Okay, so, you know, what we try to do on the show is, is, you know, give people an idea, like, what, you know, everybody sees like different things. Oh, I can put an AI chatbot on my website. Maybe I've seen companies take AI, build like an AI that accesses their like frequently asked questions. So you could ask the AI a question instead of you scrolling through like 1,000 pages of on a document, you know, on an FAQ from like a software company, you know. So what kind of things can business owners do with, with AI to kind of dip their toe in?

Oh, there's a lot of buzz around the topic nowadays, and I fully understand. It's such a clever machine that's capable of mimicking what you do. And to be fair, there are solutions or people trying to sell solutions about pretty much anything. You can build like a second brain so you can store all of your documents and things and think about it. And question about it, you can create like support ticketing solutions and writing better emails and so on and so forth. But the thing is that although one day will be there where AI can take these responsibilities all alone, as of now, everyone should be kind of aware and cautious about leaving AI to talk alone with your customers. The thing is that even those who develop it, so I'm talking about the OpenAI guys, Ilya Sisker and so on and so forth, they don't fully understand what are the barriers and the intricacies that these models or this intelligence and how they work. So it's not uncommon for someone to see something like what happened with like PayPal. If you jump into PayPal and there's this chatbot, for emergencies and you say, I got scammed, and there's like a meme running around the Facebook today, and then the chatbot answers, oh great, so how can I help you? You see, so the thing is that we don't know exactly how to put these barriers on AI to work in a particular way. And the second thing, especially if you're doing research and you're like using it as if it was like a Google solution, so you can save time not digging on things, is that AI was not designed to challenge you. So even though if you're asking like a wrong concept of it, it tends to tell you, well, these things is kind of like that, and then like put you on a paragraph that makes you feel, well, maybe that's not that bad. So You hardly hear from any AI model something like, don't do that, Allen, that's just wrong. You should avoid it. Only if you're hitting the barriers of security and this kind of stuff, so barriers of the model. But on a daily basis, it's amazing if there's a human in my vision, the way I see it. It's amazing if there's a human taking care of it. So the way we designed AAAI, for example, is that you always have a human, like, taking action and using AI to do the heavy work. So if I'm writing an email, it will write for me, then I will prove it, and it's done.

Yeah, one of the things I've been saying about AI is people talk about, oh, well, you know, robots replace people, and then you see, like, There are robots in manufacturing and unloading ships at the, at the port and unloading trucks and things like that, you know. And AI, I think, is that people need to develop AI skills. Like today, you need to— if you have to write an email for work, right, use OpenAI, type in the email and say, hey, give this to me, make it sound a little bit better. And it's aimed at— I'm sending this email to my employees, or I'm sending this email mail out to potential customers for this product and learn how you could use it, because eventually people that know how to use AI are going to replace people who don't know how to use AI.

I, I couldn't agree more. We have like 10,000 active users nowadays, and one of the things that I spend most of my time on my YouTube channel or on support tickets is like guiding them on how to better ask things for AI. Because depending the analogy that I do, especially for business owners, which is amazing, is that I remember the time when I was on a huge board meeting room and one of these large companies that I worked for, and then out of the blue come Hookie Analyst that everyone saw as a potential. So the guy was speaking quite well. He seemed to understand everything. He was kind of very open to discussing and so on and so forth. But in the end, he lacked experience. So every time you were guiding this individual, you should be as precise as possible and guide him with context and telling him exactly what you expect out of him, because else he thinks he's too smart for you and then will deliver something that you don't expect. So the metaphor, metaphor that I use for AI is that it's a very young, skilled analyst. And if you ask it the right way, you'll get amazing results, I'm pretty sure. And you can bundle documents and contacts and stuff. But be careful because it's just an analyst. It's not skilled as you are. You are still, at least, thank God, because one day we won't be in this position anymore. But we are still the most important side of the equation. And then you have to see what I think people need to figure out.

Like, I don't know if they're— I see classes on, you know, Google has some free classes on AI, and Coursera has a class I think you have to pay for. But one of the things that I see is, is, you know, you really have to learn how to write a prompt. So I was talking to one of the people in my office, and I was like, well, you know, you don't just say, oh, I need a picture of a toolbox. You want to say, I need a picture of a toolbox on a white background. The toolbox needs to be open. We need to see a screwdriver, a wrench, pliers, a plumber's pliers.

For images, it's even harder because there are lots of variables that you should consider. But for text, in the end, if you do think a little bit and explain what you want, you get much better results. And 90% of the time, independent of the model that you're using, you don't need to have these cutting-edge models for most of the things. You'll get something that you can work on. But on the other hand, if you just get to ChatGPT and say, well, write me an email, man, it doesn't know anything. The truth is that AI is not intelligent. It has a heck of a huge memory. It learning about the whole internet, as we know.

Yeah, but if you have an account on— if you have an account on ChatGPT, every time you, you're typing something, it's learning about you, your style, what—

who you're—

what you're doing, you know.

It's, it's actually memorizing a bit about you, but the model itself stays the same. And this is the difference between humans and AI, because we can evolve And the models as of now, they are kind of static. So ChatGPT is stuck in— I don't remember the last training that they ran, but let's say December 2024. And that's what it learns and knows. Everything after that, it's kind of context. It's like adding short-term memories for it.

Yeah, I was telling somebody, I said, you can go ask ChatGPT, say, oh, This is my website, www.nationwidepaymentsystems.com. I need a content calendar for the next 30 days about this one product, whatever that product is. This is going to be for business owners. I'm going to post on Facebook, Instagram, and LinkedIn, and I need some, you know, suggestion for 2 posts every day, and it'll spit out You know, a whole thing.

It's what it does. It can't say no. It can't get to you and say, look, Allen, I'm not exactly a huge SEO guy. I don't know much about posts, and I cannot judge what's the best post for payment system instead of like a beverage company. So it tries quite eagerly to deliver something that you like, and Well, with the content that we have on internet now available, it's, it's pretty, uh, comfortable. You can be comfortable of getting something that is good, maybe better than you would do it by yourself because it takes some time, but it doesn't mean it's good. So this is why there's always like Flea that you should like keep in the back of your head. So work a little bit on your, on your results. That, that's it. Yeah, you have to take what they give you and then you have to enhance it.

So let's say you're writing a blog post and maybe doing a blog post might take you 3, 4, or 5 hours. That's over a couple of days to put together a blog post. You could start off with a paragraph, then tell it the tone and who it's for, and it will literally pop you out a blog post. And then, you know, when I do that, I take it, I put it into like Microsoft Word, and then I want it to sound more like my voice. Right?

Yeah.

And how I want it to be, the tone that I want. And also make sure you're not, you know, you don't have a bunch of, uh, complicated $10 words that the average person is not going to know what that word is.

Yeah.

And then, and then you improve upon what, what you got, you know?

That's, that's perfect. That's the way to do it.

And that's what AI is all about. It's like taking your task that would take you 4 or 5 hours to do, and maybe you can do it in an hour. Yeah, you know, and that's where I see, like, right now, things that people can do with AI.

Something— it's amazing. I have— I use— I use my tool for pretty much everything that I do now. So either if I'm, like, translating content— I do speak English quite well, but I'm not, like, native, so sometimes I miss a particular word and I have a prompt, a preset prompt that improves my English and without losing my style. I do have a sort of a persona set that every time that I have it active, it writes as if it was Felipe. So it uses the same level of vocabulary and so on and so forth. But it's exactly what you said. It's a sort of a tool and we should take it like a tool. For now at least.

Yeah, I mean, I've used it for different things, like, oh, I need to write a letter, I need to write an email. Um, I needed to write an agreement one day, and I was looking at an agreement that I had, and I said, oh, something was missing. Oh, I need something in there that says something about, um, intellectual property. Yeah. And I said, oh, in this contract it's missing a paragraph on intellectual property and it spit it out. And then you look at it, you go, okay, that sounds— Rocky, my dog, he doesn't like—

you want to talk about the AI with us?

He's like, oh, we're talking about AI. There's no AI, dog. Rocky, come here, come here.

No, I use it for this and also especially for contracts. I hate reading contracts in I don't know how is law in the US, but here in Brazil they're kind of repetitive and stuff. So I can throw in a contract and say, what are the parties involved? What are the key things that I should be aware? What are the risks for each of the parties? What are the cancellations? Um, how is cancellation of this contract defined? And then I can read. And most of the times what happens is that I get to our lawyer and say, well, I know it said this, this, and that, and this um, it's kind of worrisome because of this clause and so on and so forth. So it's, it's easier for me not to be a specialist because ChatGPT is not a lawyer and Triple AI is not a lawyer also, but it can guide me on learning a little bit about it and like, uh, asking better questions that the time that I need to do so.

Yeah, especially like, yeah, what any type of agreement you can put it in there, you can say, oh, um, you know, tell me, you know, what are, what are the, you know, the term of the agreement? Is there anything, any paragraphs contradicting each other? Things like that. And then it'll highlight that for you. And especially if somebody sends you a 20 or 30 page agreement, I mean, that's a lot to— it's a lot to, uh, you know, to sit there and read yourself. And you don't want to pay an attorney $200 or $300 an hour right, to read a 50-page agreement. If you find a few things in there yourself, you know, that's going to save everybody a lot of time because you can go back to the person and go, hey, I saw this, I saw that, I saw this, is, you know, is this negotiable, is this not negotiable, right?

And that's it, that, that's the way to go.

But I think today, you know, some people are like getting content from ChatGPT or any of these other ones and they're just copy and paste it and put it on their website. And I think that that's kind of not a good idea.

I'm into software and I know like lots of sketchy practices of people all around. So let's just share an example with your audience. So everybody knows that SEO is an important thing. So if you're like well positioned on your Google searches, you'll be like doing better with your business. Someone came up with the idea of building massive content for blogs to rank better. I think this is somehow an awful practice, but it's just what it is. Most of the blog posts that you see nowadays are not written by humans and they do not rank because they're good. They rank because there's someone spitting 1,000 blog posts an hour. Using ChatGPT or OpenAI. Claude is better for that, using Claude APIs to generate this content and adding pictures and goes down to the level of like locations and geolocations of searches. So if you search something like a plumber near Fort Lauderdale, chances are you hit one of these blog posts. Made for you to get this as a first result of Google. Um, it's just what it is. It's, it's a tool people will use for good things, some will use for bad things, and we got to accommodate and learn about it and take the best of it following our ethics.

Yeah, we use it for— we, we actually use a, like, an SEO tool that— so you can tell it, you can give it a page And then it'll tell you like, oh, here's the words you should meta tag. Here's the hashtags you should put at the bottom. Here's the—

But you're still crafting the blog post and the page and the results. So it's different from pushing a button and saying, well, now hit me with 5,000 blog posts to cover the whole Florida regions or counties. With good SEO results. And then without even reading that, you push it to WordPress and you say, well, having only one blog, I can't achieve a good, uh, SEO result, so I should have 10 blogs posting 5,000 articles to cover Florida. And there, actually, so I actually saw somebody that had a website like that.

Yeah. And for some reason, I don't know, Google like found it and they were like not happy about it.

Yeah, there's always this discussion about like Google hunting people like doing these black hat things on SEO and they should do it because they are under threat now because OpenAI built a search engine to fight Google. So in a sense, they should deliver better results every day and filter these things out. But they say they don't care that much about AI content as long as it delivers the results for the users.

So yeah, I can't really tell. Like, Google seems to be like very up and down about, you know, what the search engine results— one of the problems I have with Google is, and even Facebook and, you know, anything else, you know, mostly just Google and Facebook, because, you know, any place where you have to put an ad, right, they don't have a way for me as a business owner, if I want to have an ad that's only targeted to South Florida or only to Florida, right, I have to pay the same as a huge billion-dollar company, right?

Yeah.

And they haven't figured out how to do that. Whereas if you go back to the old days— we're both older, so, you know, you had the Yellow Pages Yeah, yeah, papers, right? And if you had the newspaper and let's say you owned a restaurant, didn't matter whether you were a chain restaurant or a restaurant that did, you know, millions of dollars a month or a small restaurant, you could still buy an ad in the newspaper. You might not get a full-page ad, but you could get a quarter-page ad. And we don't have that. Small businesses don't have that opportunity on Google or Facebook to, to get that, you know, to get that local feel. You know, I hear, you know, we hear complaints from Congress and this and that, and then the EU, and they talk all about this stuff. But, you know, you know, it's all pay to play now to get to drive traffic.

I don't think, I don't think they ever thought they would achieve this size because in the beginning, uh, right Putting your ads online was a solution for that because large companies would have radio ads and TV ads and all of these huge events, but the small business guys like you could use Google to localize and put your ads in a good fashion targeted to your specific audience in a region, but then came up that everyone decided to use internet for pretty much everything, and all the giants move it to the internet ads. And now Google, of course, looks and say, well, I'm going to care about the large guys that I have on my portfolio or the smaller ones. But look, they don't have much to do else than fixing their search engine and their ads because the competition is like crushing them. The truth is that Google in AI is far behind OpenAI and even behind the open source models. What they do have is a huge ecosystem that OpenAI doesn't have. So they have Androids and they have like App Store and they have Chromebooks, which I do know that is quite interesting business in the US, not that much in Brazil, but they, they hold this like final connection with the user. And hopefully, and I believe so, they sometime will catch up and, and bring something interesting to the table. As of now, they're just losers, and it's sad to see. Yeah, yeah.

Like now on the, on the Google, when you run a Google search, you get at the top like a generative AI result. Also, you know, they'll give you something, although it doesn't give you any citations of where they got this information from.

You know, it's hard for them because it goes straight to their business model. Because if you, if you got to put a summary of the results before the ads, chances are that people will stuck there and don't move to the ads or just ignore the ads.

I just wish they had a— I wish they just had a local, like you say, oh, I'm a local business, I want to advertise and they had a way for you to do that, but they don't.

When we, when we end this conversation, I'll call some guy and ask him to solve this for you, to solve the problem.

Yeah. So what do you think is like— here's things with AI I see people doing: chatbots on websites. You think it's good or bad?

I think it's usually bad, um, but because It's kind of hard to understand because it's very— I don't know the word in English, but it's very easy to fall in love with AI. If you're like, let's say, my father, if he jumps into a computer and starts talking with ChatGPT, he probably thought that that was kind of like magic because the answers are pretty good and and the vocabulary is fantastic, and you can ask in Portuguese or in English and it goes the very same way. So it's easy to believe that it's good results. So what people came up with is this, like, chatbot solutions for, like, pop-ups that you can put on your websites. And the only thing I think a business owner should be aware is that there's absolutely no way for you to be sure that the answers will be correct. If you use that for lead generation, if you're going to sell your products and you put generic information about the products that could be asked for this chatbot, and if you're trying to catch someone's emails for newsletter or mailing lists, that's good, it works. People are used to that. But if you try to use that for something more interesting, like real problems that your users are facing, billing issues or this stuff, just don't do it. Because if, let's say, you know what are some of the easiest ways to break a chatbot? Is to ask the same thing many times or changing slightly the question. Even though if it's prohibited, at one point after 50 questions, it will deliver the answer to you. Think this cannot only turn into a problem because a user could print out an answer from the chatbot and send to you and say, well, you gave me 90% discount and I'm trying to use the code and you don't allow me to do that. But depending on how you train this chatbot, there could be licenses or keys or something that should not be shared and will eventually get shared. Think of it as a black box. So for foolish things, go ahead, use it. But for serious things, it doesn't cost that much to read again and understand and try to like wrap up better the solution, especially when you're talking with clients. That's the key.

So as a customer, I don't really like it. You know, an AI chatbot I'll interact with it. But I'm gonna ask like, oh, I need a real person and try to get—

Yeah, we suffer this in telephones with the URAs. So every time we call it like American Express and press 1 for this, press 2 for that, press 3 for that.

That's not— that's really like, that's more like a lot of people confuse AI with automation. That's more like automation. But I've gotten phone calls like from an AI before. And the AI is like, Hi, this is blah blah blah from Florida Window Company or something. And then, you know, I'll ask it like a stupid question, like really stupid, and then they'll hang up the phone on you most of the time. But sometimes I've caught them where the person didn't really do good programming and the thing is confused and it starts saying other crazy, you know, start saying stuff to you.

It's getting better. The voice, the voice generation is getting much better now. We have like ElevenLabs, which is a company that is doing like amazing job. And I saw a couple of like video transcriptions out of their service. So they're not only doing the translation to another language, but also setting the like matching the movements of the mouth with the sound. So it's amazing stuff that is coming and it will eventually come, but it doesn't change the fact that the core of it, which is the message and what you're telling, is just uncertain. And we like certainty in business, at least I do. So there's kind of risk in new business or like how to achieve better things, but you don't want to risk with your customers.

That's it. Yeah, I think that, you know, I think people need to like experiment with it. They, you know, see what they can do with it. But I think implementing a chatbot is not as easy as people think it is. It's not like you just stick it on your website and then it's all of a sudden going to read your website and know how to answer everything.

One of the things I use all the time is I have this full documentation of Triplo AI. I have a mind which I call, which holds all of this content. I also extract this from FAQ from marketplaces that I have been selling. I have this huge base of content about my tool and whenever someone comes with a new ticket, I don't need to rethink that much about it. I can jump into the question, bring Triple to the screen, say, answer this guy using this, this and that. So it's just like a very simple short sentence that brings the perfect answer for this particular case. Since I know these things, I can hit send after that. It's more for mundane users, mundane uses on my cases now, but those that takes time.

So if people want to experiment with AI, what do you think is the best kind of things for them to try out, like blog posts and content creation and things?

Well, I, I think, I think that ChatGPT is for sure the best entrance on this. New tool, just practice. It's very good, especially if you're researching. If you don't know exactly what you want, you should use ChatGPT. I do use that if I'm like starting a brainstorm, if I want to explore different things. This, it's a very good way to start. Then would be the—

who would be the perfect kind of customer that would use your product?

Oh, usually solopreneurs. People that is on a race against time. Because the concept that I have on AAAI and the idea behind it is that it's not like a cannon. I don't want to explode everything. I want an arrow to do a very precise thing with one shot. So whatever I'm like working, if I'm answering a support ticket from a Spanish guy, although I do speak Spanish, I don't want to spend time with that thinking and rethinking, remember the vocabulary, so on and so forth. So, and I don't I don't want to leave the support ticketing system that I do use. So with a single hotkey, I can translate everything using AI, which is much better than any other tool. Uh, when I'm answering support tickets, I already shared that with you. I have a process where I have to create like Pipedrive cards for business, uh, that for leads that I'm working on, and I don't want to jump into their sites and like extracting what they do and getting contacts and describing and so on. So I have an automation for that that I trigger out of Tripwire AI. So it's, it's always about those small things that I would have to jump into Google and search for an Excel or Google Sheet formula, and I don't want to do that anymore. And I can't do that anymore because it's kind of addictive. It's, it's an assistant. It's always there. So if I hit Ctrl+Space now, I see it over your face and I can drag it around and I can ask anything. I can ask, what's the weather in Fort Lauderdale now? And then it will answer me. And I could, for example, in a meeting, I already did that. I asked it questions about topics or acronyms or things that I didn't know. So I could jump straight ahead and say, well, look, if you're considering the COGS of that product, you should— I do know what's the COGS of cost of consumer goods. And these are the things that AAA does quite well. It's like a one-shot thing with predefined prompts and minds and automation and personas that you can set that it will always talk about the Brazilian Felipe or the Felipe speaking Spanish or someone that knows quite a lot about Python and have automation set to deliver results to other things.

I didn't even think about, I guess it could translate things into different languages. I never thought about it that way.

It's amazing. It's amazing because the prompt I use, it respects the originalities and slangs and this stuff. As a Brazilian, if I do speak fast or if I speak normally, Of course I will have some slangs here. When I translate this to English, I lose all these slangs. Translating this way using AI, it tries to keep the context of these slangs and use local slang. I have, for example, we do have more than 200 smart prompts for that, which is a prompt for translation. You can ask for it to translate to French as speaking French as spoken in France, or in Switzerland, or in Guyana, because they're like differences between the way people speak. So it's very, very good for this.

So if there's one piece of advice you could give a business owner about how to dip their toe into AI, what would you tell them?

I would tell, and I told this quite often, I say this quite often, there's no way out. You got to learn about AI because if it is like as it is now, in 2 years there will be no way of not working with it. You'll be behind your competitors if you don't have it. On the other hand, I'll reinforce and I don't want to be that one repeating the things, Just be careful, especially when you're talking outside. I still believe that we are at a point where humans are still much better and more intelligent than any AI. No matter if it's GPT-3, which is the latest from OpenAI, we are still much better than it. But it demands you some time and you should like review everything that it's spitting to be sure that it's good.. So it's better for those who know what they want. If you don't know, if you don't know what you want, hire a specialist. That, that's the thing.

I'd say learn how to write prompts, and that's it, and experiment with that. So if people want to find your product, what's the website?

Oh, that's Triplo AI. Triplo is like triple in English with O, so it's, it's a Latin word for 3 times. AI is the extension of these domains. AAAI, I think you'll love it. If you have any questions, don't hesitate on dropping me a message on support. I'm always there and I'm happy to help anyone learning more and doing more with AI.

If you're looking to learn more about AI, hope you enjoyed today's podcast. Visit Felipe's website, check it out. I don't— and, um, you know, and start experimenting with AI. So thank you for listening to B2B Ball. Thank you for being on the podcast today, far away from Brazil, and talking about AI with everybody. I think this was a pretty interesting discussion, and I hope everybody enjoyed it. So if you like the podcast, share, like, and subscribe. Everybody, peace out and carpe diem.