Latest / Selling AI / AI Just Accelerates Your Inconsistency
Transcript
- Warren Kucker: Hey there and welcome to another episode of Selling AI. Today we're diving into how prospecting is changing with the technological revolution we're experiencing with LLMs. I have Marvin Martinez on the show today. Marvin is the founder of Bansaw.ai, a consultancy that helps small and mid-sized businesses plug LLMs into their sales and marketing processes and their existing tech stacks. Here's what makes this conversation interesting. Marvin spent 13 years running different operational teams before founding Bansaw. So he's coming at this from an operator lens, not necessarily like an AI evangelist one. His core thesis is that if your sales process doesn't work today, layering on new technology isn't going to solve any problems, it's going to create new ones. And so in today's episode, we're going to talk about how, like where are the right wedges in the go-to-market process to automate. why the human connection has gotten even more valuable now, and where he draws the line and actually his own prospecting between like automating things and a genuine human touch. All right, let's do it. Hey Marvin, thanks so much for being on the show here with us today and talk about how prospecting is changing. Welcome to the show. Marvin J Martinez: Thank you so much for having me this day. It's a pleasure. Warren Kucker: That's great. I'm looking forward to diving in. So first, just give us a bit of your background, your career arc. Tell us about your 10 plus years in operations and then how it kind of led you to a path ⁓ with what you're doing with Bandsaw. Marvin J Martinez: The way I like to talk about my background and experience is that I am not a person who knew AI and then decided to help business owners. I'm exactly the opposite. I'm the guy who was in operations and I just happened to learn AI after the AI boom that happened a couple of years back. I spent over 13 years working in operations in a business process outsourcing company. I was kind of the general manager of the company and I was in charge of managing the entire process of the company from the recruitment side going all the way to training, quality. operations, IT, HR, and my main job was to ensure that everyone was running smoothly. So part of the skills that I developed in this role is to map the processes, do workflows, diagrams, ASOPs. And when the AI hype started, you know, I came to realize there's a lot of the things that I do every day that I do repeatedly that can be actually automated using AI. And that's when everything clicked and I decided to co-found ⁓ Bansal AI. Right now, through this company, what I do is to help other people like me that they know how to run operations, but that they are struggling to either ⁓ grow or they are spending too much time in front of a computer. So that's exactly why I helped. And that's kind of my background and the reason why I stepped into AI and automation. Warren Kucker: Well, let's focus on the growth piece. What in particular are you finding is super helpful when you're working your engagements and helping companies try to push revenue up? Marvin J Martinez: One of the things that I'm finding very consistently is that business owners are very good at what they do. Founders are very good at what they do and they hire people who are extraordinary in the processes. Sales, customer service, marketing, so on and so forth. But they are spending too much time doing the same things. And when I come into the conversation with these people, one of the things that I found very repeatedly is that they know how to do things, but they don't necessarily have have it mapped. They don't have an SOP. So when I talk to them and I tell them, I can help you save time because I don't want you to spend the entire Friday afternoon looking at a computer doing your thing. But when we talk about the process itself and because they can't explain it. then that's where the friction starts. And that's what I've been seeing happening, happening, happening very repeatedly. I feel that business owners, start with the tool first. They sometimes think about, hey, this tool is cool. I know that I can do so much stuff with it. But then they start framing the business around the tool when it should be the exact opposite. The tool needs to be Working around the business and that's where the main ⁓ issues happen when it comes to AI automation and saving time Warren Kucker: That's interesting, that's awesome. So as part of that core thesis, you talk about how you should be plugging LLMs into your existing process rather than reinventing the process from the ground up. Why do you think that is? Why do you think that approach, rather than just reimagining how the whole process should work, works better for companies that are implementing new tech? Marvin J Martinez: my point of view is that AI always amplifies your systems, ⁓ whether they good or they bad. ⁓ So what I'm saying is that if your process is not clear today, ⁓ AI will not magically fix your issues. What it will do is just accelerate inconsistency. So the companies that are getting results This is what they understand compared to the others that they don't. Number one, they know how to generate leads and where the leads are coming from and how they move from the different parts of the marketing funnel. When they know about you, when they start consuming your content, when they know that you can help them, and lastly, until they book the call with you to get the services. The companies that I know that are being successful in AI implementation, they know how to follow up with their customers. They know that when a customer comes in, you need to call them in five minutes or less because they want to buy at that moment. They feel the need. If you call them the next day, chances are that they already contacted another competitor or another company that provides the same service. And lastly, the companies that are being successful, they know who does what. So what that means is, you know, ⁓ I have a closer, I have a person that is very good at speaking, at delivering presentations. So in the funnel or in the entire sales process, they know who is responsible for which results. At that point in time, when you implement AI, then AI becomes a leverage instead of noise. Warren Kucker: That's a good way to look at it, right? You don't wanna amplify, you're gonna amplify your process. That's the negative and the positive, right? So you could end up creating bigger gaps than you had before if you're accelerating a process the way you're talking about. ⁓ That's really interesting. Okay, cool. So walk us through like a typical engagement. I know you do a varied amount of stuff in go-to-market. What's the most frequent thing you're trying to help someone with and how do you kind of get them off a blank page to, an end result process that works well for them. Marvin J Martinez: In my initial conversation with ⁓ my founders, people that I speak to, other business owners, I try to ask very important but easy questions. So the first thing I usually tell them is, show me, number one, what is eating up your time? What is keeping you up all night? And then, most importantly, show me the manual version first. Walk me through it. Show me your system. How do you connect here? How do you connect there? And then we start measuring as well the time it takes them. Because most people think that when you implement AI into your companies, everything is fixed. But they don't really understand if AI actually delivers ROI at that point or not. Because they don't have a KPI to say, this improved compared to my previous state before I implemented AI. So that's kind of my question, right? So let's talk about the manual version. How long it takes you? How many people do you need for this process? And at that point in time, I know the broken point if there is one or I know if we are ready to start. My next point in the conversation is, well, now that we know exactly how your process looks like, let's start with the easy ones, with the easy process that we can automate. Because most people or most companies, they believe that they have to do this very huge complex AI agent automation in order to look cool. in order to say, I am the best company out there. Maybe that's part of their promotion. Maybe it's not. But at the end of the day, I tell them in my conversation, hey, let's start with something simple. Look at this use case. A founder receiving 50 to 60 emails a day and spending two minutes, only two minutes, just going through it really quickly. If you do 50 emails times two minutes, that's 100 minutes, one hour and 40 minutes. of one person going through admin work that it might not be necessary. Why don't you start by just implementing an AI automation that can actually pull up the most critical and important emails on your behalf and just focusing on them? That's when you see real return of investment. And that's kind of my discussion with the founders. What part of your process ⁓ is the best to automate at this point in time? But most importantly, tell me how it looks. when you do it manually. And this applies to everything. Admin stuff, customer services stuff, and also sales stuff. Because this is where it gets very, very exciting. Warren Kucker: That's awesome, yeah. In particular, it's like anything else you've implemented over the past 15 years, right? Like what you're describing as a typical change management playbook, right? So like with, again, new technology, what's old is new. Do you find that there's anything different as part of like, again, you probably implemented for the first 10 years of your career prior to like LLMs, you implemented a lot of process change, SOPs, you analyze someone's process, you at what's happening today. you figure out where the time savings are, the barriers to success are, and then you find solutions there. Now you just have a new tool in your toolkit. But that tool is pretty drastic. Is there anything that you find that's different in this type of ⁓ analysis of the problem and solution to the problem than it has been before LLMs? Marvin J Martinez: You know, surprisingly, surprisingly not. And there's so many tools out there that can do the same function that the only thing that makes a difference is how you use it. And I am a person that really likes to talk about examples, real things, right? Not the hype that you see in social media, things like, hey, my AI, or I have an agent that can replace my entire go-to-market team in just one day, right? Like, not that type of stuff, but... Think of this, nowadays everybody has access to ChatGPT, Claude, Gemini, Grok, DeepSeek. Imagine how many LLMs people have access to, but they are not being successful. the same rate or at the same level because they use it differently. It's important that you understand your input, what information you're gonna feed the LLM or the AI with. ⁓ Like for instance, if you are going to work on a cold email outreach campaign or if you are working on social media content creation, right? It's critical that the LLM knows your company, knows your voice, knows your brand for you to be successful. So even though there's a lot of technology and tools because people use it differently, you know, you will be getting always the same results. So ⁓ in my opinion, in my point of view, what matters most is how you do things rather than the tools that you are using to accomplish those things. That's the way I see it. Warren Kucker: Gotcha. That's interesting. I think part of it too is because it seems like a more drastic change, the change management aspect of it is bigger for folks. On the flip, something that I've noticed in this space is when you do an ERP implementation, something big, or even something moderate, like a new piece of sales technology, you're switching from outreach to sales loft, there's a decent chunk of change management that needs to occur. People are used to working in their old ways. They're now logging into a new platform. They have to figure out where all the light switches are on, right? On the flip, with like Claude or OpenAI, the access to these are just so instantaneous that like you, it doesn't seem like a big lift in terms of like we have an exact functioning different way of working because you just go and do some work, right? But it makes it pretty scattered, right? So again, interesting from a change manager piece. When you're implementing these processes, let's say, specifically in go-to-market, let's say you're implementing a new outbound process, how are you seeing the change management piece work out? What's the reluctance to change? Is it the same as it's always been, or are you finding it to be of a different nature when you're looking at ⁓ automating cold email and research and outbound? Marvin J Martinez: It actually depends on the owner and how they manage that part of the implementation. And I've seen employees ⁓ being very hesitant of using new AI tools, technologies, LLMs, even though they are everywhere. But I have seen some other business owners being very successful. And what's the difference? Number one. is the level of communication that they have with their employees. Because what I've seen is employees and people in general saying, you know, AI is going to replace me in what I do. But that's not 100 % True, AI is going to replace 70, 80 up to 90 % of the process that is repeated every day, but it will never replace human touch, transparency, decision making and strategy. So the companies that are being successful in implementing AI, they start with the vision and then they send that vision over to their employees by communicating, by saying, hey, the purpose of these new AI implementation for Alba or outreach is that you guys spend one hour less doing research on a prospect before going into a sales school and then you start providing that benefit to the employees. Imagine you getting into a sales school knowing everything about the lead that you are going to speak with. That's what makes a difference in your business. Because now the lead or the prospect that is going to talk to you, they will know that you really care, up to the point of getting all this information about him using his website, LinkedIn, Google search, whatever. I had this founder speaking with their people, telling them about the benefits and telling them that AI will not replace them or will not replace the entire process because AI can hallucinate as well. So it's important to always have, I call it a human in the loop to ensure that the output of the AI, agent, automation, or whatever you have implemented, it's as accurate as possible. Warren Kucker: Awesome. All let's get really tactical. Let's say you're working with a team, has a sales director, a dozen AEs under the sales director. What are some practical ways that you feel like they could and should implement AI? Maybe one in particular. This process has changed drastically. Implement this series, do it in cloud, do it via piece of software. What's something that you see an approach that someone could get really tactical with? I'm in a sales process right now. Marvin J Martinez: Let me give you an example of something I did with a staffing firm. We built this AI agent automation for the president of the company. And this person had a very unique approach. He said, I don't want to use AI to speak with my clients, my potential clients. How can we actually implement AI in a way that makes sense for us? And then we started to walk through the process. told him, and I want to go back to the previous questions and answers, understanding what you do manually first. So ⁓ what this person explained to me is, because again, this is a staffing firm, and they said that they were ⁓ looking into job boards every day. to see what companies were hiring and for what role. So they understood the company. they made sure that the company fit their ideal customer profile. And at that point in time, they would start doing some enrichment process, meaning finding who the decision-making person was, finding their LinkedIn profile, their email, finding ⁓ their website, and so on and so forth. After getting all that information, use that to their advantage to reach out to the person via LinkedIn and also via email. different approaches, right? They call it multi-channel outreach. So that was the process that we were trying to automate because at that point we knew exactly how the manual process looked like. So I did help them to implement an AI agent automation that would monitor the job boards every day at a certain time. So every day at 7 a.m. we had an AI agent that would go through the job board and look only for new jobs, for companies that were part of the ICP of this this staffing firm. Now if identified then it was going to use Apollo IO via API and I want to like come up as to Taney call but just connecting that company via Apollo so that Apollo could extract information from the company. Decision maker, LinkedIn profile, email, everything that we needed. Just putting that into a Google Sheet By the way, you don't need to have a fancy tech stack to do things. You don't have to be using Hopspot, Salesforce. For this person, simple Google shit was the right thing at that moment. And then putting all that information, just creating an initial outreach paragraph. for the president and that the president was doing personally that outreach via LinkedIn. So that's an entire sales process where AI was doing the heavy lifting of doing the research, finding the people, but the actual communication still came from the human. This is important because I think that people confuses automation with removals. And that is not true. The highest performing systems right now are hybrid. AI handling the repetitive coordination, repetitive Warren Kucker: Gotcha. Yeah. Marvin J Martinez: and humans handling trust and judgment. So that's a winning combo. Warren Kucker: Yeah, there's a bridge across there. It's interesting. For a long time, I sent, I'd say, automated messaging out that was well written. And my thesis, and it worked really well for the most part, but thesis was that like, we're overrating how well a human generated message is written and how personal it is. And I've actually said this on a call yesterday that like, when you read something that AI generated that you're going to send to someone, whether it's an email or a cold outbound email, like to message. You're almost always going to think it's bad, mainly because it's not you, right? It's like a, it's foreign, like to how you speak and how you think and how you read. And so it seems worse, right? Like the, but in in the flip, trying to take a step back and actually evaluating, would this be a valuable piece of content for someone to receive? What is the likelihood that there's some obvious errors in there that make it invaluable? If it's less than 1%, I'm going to have the same problem. So like writing the email, I'm going to have an error rate of 3, 5, 7 percent where I mess up on that messaging. And coming out of the context, what I found though is that at least getting eyes on all your messaging as quickly as possible or actually being the one to send the message does drive better results. So I'm like getting the core of the message, but I tweak, edit, send, like myself, in a way that I know that it's not purely AI generated. It just gets me off a blank page. Are you finding the same when it comes to that type of like automated messaging? Marvin J Martinez: Yes, I have to agree to that 100%. That's exactly what I'm finding. What I'm finding, I'm sorry. You know, the more, this is the way I see it, the more AI generated noise exists, the more the authenticity extends out. And not only in cold emails or LinkedIn ⁓ outreach messages, think about social media. How many LinkedIn posts do you see out there that you can actually tell from the very beginning it's AI generated. So like for instance, ⁓ if you see a post that start with the word must, M-O-S-T, that's AI generated. Right? If you, yes, if you notice like contriving ideas, like for instance, this is how I save money, not by not spending less money, not by, you know, buying ⁓ like cheaper stuff, but by saving money, you know, those contrarian ideas are AI, like you can instantly tell if something is AI generated. So what I've seen that people respond to better is the context, the proof that someone actually understands them. And actually that human touch where people is talking, there are some tools that you can actually use in your automation that can add a little bit more of transparency. For instance, I use SendSpark to create ⁓ custom videos for people. If I have a list of 50 people that I can reach out to, I use SendSpark to create one custom video for each one of them. But again, the message when you send it, like when it was instantly created by AI and you send it right away, know, it's not gonna work. You will not get the same level of reply rates. ⁓ At the end of the day, I think that when everybody automates the same way, authenticity, that becomes the differentiator. And that's what we should work on. Warren Kucker: Cool. How do you measure the results of these programs? What are you looking at to see effectiveness? Do you see effectiveness drop because we're spending less time on it and worth it? ⁓ What's that process like? Marvin J Martinez: There's two ways to looking at it. ⁓ So the way I see it is, you, let's talk about a cold email outreach campaign. You send a thousand emails in one day and you get a 1 % reply rate. One, two, three, that's, let's say kind of the same or the standard. So if you spent three hours less, doing the same structure, the same emails, the same software, and still getting the same results, that's a win, because you're spending less time. That's one way of measuring it. If you today are spending less time researching, structuring, creating the message, drafting the email, then that's a win. Now the other way to look at it is you still spend the same time, but you are getting more replies. because you are personalizing, because you are ⁓ using AB ⁓ variations and everything. So that's a different way of looking at it. When I work with companies, I focus my ROI in either time save or hour save or errors that were reduced. But the actual results of ⁓ leads coming in, how many people you convert after your booking and everything. I typically leave that up to the company based on their processes. Because again, going to question one, if your process, if your self process is not working today, it doesn't matter how much AI you add to it. you're in, let me go back at my beginnings when I was doing Banzo AI. I started to do cold email outreach and getting no responses. because I was sending my outreach to the wrong people. So if my ICP was not defined, if my super consumer is not defined, it doesn't matter how much AI I add to it. And when I speak with founders and companies, that's what I tell them. What is your current ⁓ book rate? What is your current conversion rate? Because I will not focus my results on that sales KPIs because that's not what I'm targeting. I'm targeting the systematic process of researching adding the repetitive stuff. So that's the way I see it. Warren Kucker: way to look at it. Couldn't agree more. Again, same scenario. Knowing your customer, knowing your personas, going after the right folks, using messaging that works. Some of the baseline research has gotten easier, so then the importance of authenticity is harder, if I was going to summarize that up the right way. ⁓ Cool. Well, maybe last question. ⁓ Give me a mini hot take. Where do you think the technology will evolve over the next year two or three when it comes to... Automating processes or how we buy what do you think is gonna be could be drastically different if you're trying to look into a crystal ball Marvin J Martinez: Well, I'm gonna give you an opinion that might not be too popular. ⁓ What I think is that with technology, the AI tools will evolve much faster and it's gonna be way better over time. One, two, three, like the tools will become monster. They will evolve much faster than buyer behavior. I can guarantee you that. We're gonna have better agents, better automations, better workflows, more people creating fancy stuff because it's going to be way more accessible to people that are not technical. But the truth is that even when people know all that, one thing that probably not everyone will be focused on, will focus on is the importance of having ⁓ like a closer, a salesperson in that entire pipeline. So for me, the winners over the next five or six years are not the companies with the most AI implemented in their companies. They are the companies that will combine AI efficiency with a strong human positioning. So that's what I think is going to make a difference. Warren Kucker: I love that combination. That is great. Awesome. Marvin, where can people find you? You writing anywhere? Where you posting? Are you going to conferences? If people want to connect with you, what's best for you? Marvin J Martinez: Yeah, so I typically post only on LinkedIn. ⁓ Just by looking Banzo AI or Marvin at Banzo AI, they can find me or they can also look at my unpopular opinions in my website, Banzo AI. I have some podcast episodes recording right there where I talk about the real use case scenarios ⁓ when AI is actually implemented. Warren Kucker: That's awesome to hear. Great. Well, Marvin, it's been a pleasure. I'm absolutely looking forward to seeing where you continue to push the market in this types of automation. And we'll see if your prediction comes true in the next couple of years. Marvin J Martinez: We'll see. Thank you so much for having me once again, Warren. Warren Kucker: Awesome.