Host: Hi everyone, welcome. I am super excited to be here. Today we have a phenomenal presenter coming up, we have Mario Peshev. He is a global business advisor, he is also the CEO of DevriX and Growth Shuttle, both of which specialize, I believe, in scaling SMBs. He is also the author of the bestseller MBA Disrupted, and he’s an angel investor, I think investing in over 18 startups, which is amazing. So welcome to the stage, Mario. He has a wonderful presentation prepared covering AI-powered marketing, navigating the future of customer engagement. So I’ll turn it over to you, take it away.

Mario Peshev: Hey Shay, thanks for the intro. Hey everyone, hope you enjoyed the previous presentation. I took a look at what Andy has to say, and all I want to say is he’s been a great inspiration for all of us in the space who have been around for the past two decades or so. So I’m pretty pleased to be speaking here at the Prepathon, and kudos to DigitalOcean, Cloudways, and the entire broader community for making it happen.

So today’s presentation is about AI-powered marketing. We’re going to talk about customer engagement and different ways to incorporate and implement AI in your day-to-day to serve different purposes for the broader B2B community, and it’s also applicable to B2C and D2C. Prior to that, just a brief warning, I’ve got two kids at home and a dog that’s going through a long week, so if there are any noises behind the scenes, apologies ahead of time. We’re living in a semi-remote world, so some surprises happen every now and then. Without further ado, I’m going to move forward with screen share. Shay, I can use your help if there are any specific questions in between, like “what is this tool doing” or “what else can we do,” I can do some behind the scenes. We can also do a live demo if that’s of interest. I’ll keep watch on the chat, so feel free to post your questions.

Host: That’s amazing, yeah, drop your questions in the chat everyone, and let’s dive right in.

Mario Peshev: So a brief intro. Shay was kind enough to present some of the accolades. I’ve been working with over 400 different executives as a business advisor. DevriX, our marketing and WordPress consultancy, has scaled 10 billion sessions last year, 1.65 billion in total sales with affiliate and GMV. And I’ve been around the block a couple of times over the past 20-something years. One thing I have at the bottom of the presentation, I’m not an AI guru. A lot of people on X say, hey, if you’re not designing under the hood in OpenAI, you’re not an AI dude, that’s fine. I’ve done machine learning 12 years ago for banks, so I’ve got some technical chops of actually building specific AI models from back before it was cool, but I don’t consider myself a super-duper expert in today’s day and age, so just heads up.

Now, why have I placed a bunch of different photos over here? This is a Dreamwave presentation. Dreamwave AI is an AI to create photographic photos, so this is essentially not me, not one of these photos is myself, but I just put my photo in and it created gazillion different combinations that are pretty suitable for different scenarios. Why is that important? In reality, lots of us have decent photos, not necessarily top-notch. This is from a Forbes photo session I had a few years back, but I only got one photo back then. I needed different variations, different colors, different scenarios, so I used AI to populate that and use it in different scenarios. So using the power of AI, we can enrich social media, we can do guest photos and presentations, and avoid going with 70 different suits and shirts to get as many photos as possible. That’s one of the quick tips we’re going to start with.

Now, what are we talking about? We are discussing AI, we are discussing marketing, and more importantly, we are discussing customer engagement. When I talk about customer engagement with our clients, both the agency one and the advisory, I always touch on the fact that the modern web is all about omnichannel marketing. Why? Because back in the days, 50, 60, 70 years ago, when the internet was not a thing, we had television, radio, newspapers, and that was pretty much everything going on when it comes to outreach. The only medium was snail mail or cold calling, which of course happens at the office if you work at a place that owns a telephone, which in the early years was not even that common. So we started with a very limited number of channels, to a place where we have websites and mediums like LinkedIn and Quora and virtual blogs and forums, a bunch of different social media channels, Facebook, Instagram, Twitter, well now X, and LinkedIn and TikTok and Snapchat, different forms of emails, newsletters, various ways to connect with people through their corresponding best channels, PR outlets, favorite media sources, and podcasts, and the list goes on and on. So capturing customer engagement is nearly impossible unless we engage with omnichannel marketing. We want to be omnipresent. The best way to be successful in business is to be at the right place at the right time, and the only way to do that is trying to cover as many grounds as possible all the time. And that’s essentially why we need AI.

One thing I want to touch on, regardless of whether you’re dealing with paid ads or social media or sites, affiliate marketplace deal pages, or anything else, ultimately communication mediums over the past 100 years have been about these four: text, images, audio, video. So whatever you do, you want to cover as many of these as possible and then get involved in repurposing that for as many other channels as possible. When you understand the first principles, customer engagement is about omnichannel marketing, you want to be everywhere at all times, the main communication mediums are these four, you want to create this at scale and publish it in as many places as possible.

Starting with your main property, your company website, your own blog, whatever it is, what we want to do is produce elegant content, landing pages, topics, and books. Outlining my tools as we go, I use Claude and Gemini for summaries and research, that includes content outlines, proposals, email summaries, and more. If you don’t know what email summaries are, you can hook Gemini, which is Google’s model, into Gmail, you upgrade and then it creates summaries out of different conversations. Just last week we had a leadership thread with 46 different emails with roadmap conversations for 2025, lots of new ideas, and it was pretty tough to consolidate. I tapped into Gemini at the side icon of my email and said, hey, sum this up, gather the top 20 ideas, summarize them, and I got the list and pushed it as OKRs for the next quarter.

Now I’m going to open a quick separate bracket. I’m not talking about producing website content at scale, I’m not talking about programmatic SEO, we’re not talking about producing AI-regurgitated content to pollute the web. That’s not what it’s about. It is about consolidating and aggregating and combining data from different sources as quickly as possible to get inspiration and infuse the copy with ideas, to pull some stats and quotes, and then go and cut it shorter, wrap it up, add everything you need to add, and push forward. So it’s about optimization, a better process, faster research more than anything. There are different schools of thought here, so I’m not going to dive in, but you want to follow a process that’s consistent and congruent with what you want to achieve. You have to cover as many grounds as possible, your website is one of these grounds. So use the smartness of new innovative AI to drive inspiration. You can go to ChatGPT and say, hey, I have this blog, give me inspiration, find five competitors for me, give me inspiration for top-of-the-funnel keywords, give me 12 ideas of lead generation funnels I can use to pipe into my email newsletter, give me five ideas of email courses I can create to further enrich my leads once they’re in. So I’m talking about expanding your ideas, not just creating blank copy, a straightforward 1500-word blog post. I’m talking about the full funnel of ideas, content pillars, main topics, email courses, checklists, spreadsheets, everything else, all the summaries and research with AI.

Second one is chatbots and digital avatars. When it comes to customer engagement, we need to chat with our customers, provide as much information as possible. What I’ve been doing lately, and I started maybe in March when I gave an AI talk on building an AI assistant for a blog, is I built that assistant, I trained all my data, my blog, my book, pretty much everything, and just placed it on my website. So I went on stage and said, for everyone who’s interested, if you have follow-up questions and want to ask a bunch of different questions, I’m going to drop a link to my bot, and you can ask it whatever you want. It’s not going to be a sales pitch. If you want to work with me that’s fine, but if you want to learn more from me and understand how I think and how I structure stuff, just go and chat to my bot, gather this inspiration and ideas for free, and reach out if you’re interested. So website chatbots can answer frequently asked questions, provide personalized recommendations, streamline customer service, act as a virtual assistant, and even do personalized marketing campaigns. Because digital avatars and chatbots can be trained on everything you’ve previously published online, including videos and interviews and TV presence, and they start to speak your language, they get enriched by additional information based on what you teach them.

In the next slide I have the demo, let me play it and let me know if you can hear it.

[Demo audio:] Welcome to my website. I’m Mario’s AI bot, using an AI avatar to help you bridge the gap between the presentation and the post-conference experience. Navigate to the bottom right corner, there are two chatbots available, a text one trained on my content, and the video avatar who studied MBA Disrupted, my bestselling Amazon book.

Mario Peshev: So first and foremost, that’s creepy as hell, and one of my kids says, hey, where’s the creepy dot, can you show me the creepy dot again. But bottom line, now it is possible to do that. I’m actually going to do a quick demo here on my website. So what you can do nowadays is train an avatar. This is the chatbot, and now there’s an avatar that you can start asking questions. This avatar has been trained on my book, so you can ask stuff such as what is the book all about. So if anyone has any questions, let me click, “what is MBA Disrupted about,” which is my bestseller that Shay mentioned at the beginning. So this is going through the resources: “One-plus digital businesses is a masterclass in modern entrepreneurship that provides actionable strategies to empower growth executives and tech…” I’m going to stop it right here. But the reason I’m showing that is you can build that pretty quickly.

So I’m going to take it a step back. The photo is not mine, it’s something provided with Dreamwave. I uploaded a bunch of my photos, it created it, so I’m using it as an avatar, it’s my profile picture everywhere, I think it looks better than me, so I’m just using it. I passed the avatar to D-ID, which is another tool, essentially it’s called D-ID, for AI-generated video creation. So you pass your avatar out there and it creates a persona that is able to speak with a lot of natural user interface improvements, using AI to develop your persona and create a virtual avatar of yourself. And I’ve used a ChatGPT-trained assistant that was fed my book, I believe it was called Voc, so that it only answers questions on content that I have produced myself. So yes, I see Sebastian said that’s super spooky, yes it is, it’s creeping me out myself, I tend not to give this talk in the evening, not opening my chatbots in the evening because it’s kind of crazy. But as you can imagine, this is going to get better over time, it’s going to get smarter, more intuitive, more native.

I see Hani is mentioning, I wouldn’t buy whatever he was selling, yeah, I get it, it’s creepy, I’m going to run away for sure. What I can say is that my advisory stuff starts at 1,800 bucks a month, and most people reaching out, I get like 40-plus pitches a week, I say, unfortunately I don’t have time for everyone and don’t want to charge you for everything, so first, buy my book if you want, second, chat with my bot if you want, it’s for free or costs like three bucks. So this is the easiest way to draw inspiration and get to me, and if your company really needs my services, great, otherwise I’m trying to provide as much data and information for free.

Bottom line, moving back to digital avatars, there’s a lot we can do nowadays. We’re used to talking about chatbots such as the bottom-right text box that you type a question into from an FAQ and expect a direct answer, or maybe the ChatGPT-infused bots that are slightly smarter and go through a knowledge base. But nowadays we’re in a realm of digital avatars, and we see more and more of these personalized experiences showing up. Because a lot of our work is go-to-market strategies and growth hacking, what we do is combine some of these sources: go to an event with your partners, give a talk, move the conversation back to this chat, get some leads to go on a webinar to pitch something, go through the email, then move back to the chat. So an omnichannel experience also means moving from one step to another and cross-referencing that, so you can have more and more touchpoints with your customers. That’s why we’re exploring different tech and different widgets, to make sure we can do as much as possible on the digital front.

Now another two that I really love. It’s called Postwise. Postwise is one of these AI writers and this is the best one I’m using, primarily for X content, and some of my teammates use it for LinkedIn. I’m pretty picky when it comes to LinkedIn, but since I have to post like three times a day on X and still maintain conversations, this is the fastest way to spend three hours and produce over a hundred tweets of high quality thanks to great suggestions and ideas, producing viral threads based on contextual research and data gathering. The way I use this, I have maybe eight different X channels on different topics, some of the other businesses I invest in or our side brands, and all of these are focused on different conversations. Each is segmented on a specific topic, user experience, SEO, technical development, different conversations, and we’re posting a ton of high-quality content to track engagement. So we use it as a validation feedback channel. Instead of spending forever crafting unique content from the get-go, we spend a little time producing pretty decent content that’s easy to edit and schedule. Disclaimer, that’s not an affiliate talk, I’m not pitching affiliate links, I tried to invest in this one, they weren’t raising at the time, the moment I saw it I gave it a shot, ran two different prompts, and bought the annual plan, so it’s this good. So this is my preferred text generation and feedback loop channel, my assistant on content ideas. If you think of Google Suggest, where you type something and see what it autocompletes to come up with ideas, or Google Trends, this is what I’m using Postwise for when it comes to content for specific topics.

For video content, back in the day, 2016, 2017, when I started with video and posting more actively on YouTube, I recorded an hour-long video then passed it to my team to cut some shorts manually, it was taking forever. I was going to Rev.com and passing on the video to get manually typed captions for about a dollar a minute, so it was about 60 bucks per video, which wasn’t great. Then we worked on manual drafts to create blog posts and social snippets, and essentially from one video we were creating two weeks of content, but it was also 40, 50, 60, 70 hours of work. This was 2016, before AI. Nowadays, I still do the long-form video, I started a podcast as well, which is valuable, but we use tools like, in this case this is Opus Clip, where I upload the long-form video and it’s smart enough to pull out insightful moments from the conversation that I can trim down and create different lens-sized videos for shorts. So nowadays I spend five hours a month or so recording, and then I end up with 40, 50, 60 different shorts to repurpose on YouTube, Instagram, LinkedIn, which now has a TikTok-type video feed, and TikTok, just cross-promotion and cross-publishing. Additionally, for the podcast, I use the shorts to lure traffic back in. I just had a conversation with someone who has 1.6 million subscribers, and I asked, what do you use the shorts for, and he said, well, I use Opus actively to do all the shorts because I have over 400 episodes and the shorts are the fastest way to return people to earlier episodes. You can always capture 30-second, one-minute, two-minute snippets into shorts and then link back to the full episode.

Another thing I’d like to explore is how I can use AI for customer engagement through making the lives of our customers better across all the businesses in our portfolio. We’re at a stage where DIY site builders are advancing and getting better, and because this is a WordPress conference, DevriX is a WordPress contributor since 3.7, in addition to what we do purely on the WordPress front and implementing some OpenAI calls for content and product descriptions, we also use additional builders such as Bubble or Airtable or low-code stuff. Even our QA team and other people, not just our top engineers, are using app builders with AI integrations and code to build applications for lead generation, reporting, and providing additional insights.

I posted this one, this is an idea you can take a look at after: this is how I built an AI app in one hour. I used VS Code, one of the most traditional IDEs, with a Claude extension, and spent like three dollars of credits to build a persona generator for your ideal customer profile. So you type in the business description, let’s say you’re closing on a client and want to learn more about them, click generate personas, it comes up with ideas for personas, you click okay, this is the persona, and then you go deeper and say, hey, run a SWOT analysis of this business, what are the strengths, weaknesses, opportunities, threats, or come up with the main pain points, what are they struggling with the most, or use these to generate email headlines for outreach, for an email course or newsletter. So it’s a collection of prompts going one after the other that you can now type in with low-code or no-code thanks to different app builders, or by building it with VS Code from scratch using Claude, my favorite model, but you can also use Gemini or ChatGPT.

The reason I’m sharing that: back in the day the traditional marketing funnel was run ads or do SEO to long-form content, make sure that long-form content ends with an email signup or downloading a checklist or ebook. So you’re asking people to go do more work. This model is tried and tested, it’s getting more expensive, delivering less value. But now with AI and all of these builders, you can end up building a tool that solves a very specific problem for your audience. You can build a calculator, hey, calculate the average marketing spend I need if you’re a PPC agency, or the monthly maintenance cost if you’re doing WordPress support services depending on traffic, number of pages. So it’s a drag-and-drop tool that’s really easy to play with, and you get a quote and a better sense of the business. Or, type in the business, get competitor research, how am I stacking up with my competitors. So you can empower your marketing team, or if you’re a solopreneur or small business you can do it yourself, really solving problems that you no longer have to pay three months for an MVP, you can build the prototype yourself, even if it’s just a one-pager, a proof of concept to grab an email or get a conversation. I see Hani in the chat says, personally this is the most exciting part of the AI era, building apps and tools. Yes, I do agree, it’s solving an actual problem with little friction, even if it’s not super scalable, this is not a $100 million SaaS you’re looking at here, but it is solving the problem of customers in real time.

Something else we’re doing, since we’re running a lot of traffic to our website from SEO, social, paid, partnerships, affiliates, events, we have to handle all that traffic. So one of the innovative things out there is a few very specific tools came up this year, some of them are RB2B and Sidestep by Keno, and I think Instantly built one too, and there are others. These tools let you identify people on your website in real time, they give you an idea of about 20% of your US traffic and can say who that person is, give you a link to their LinkedIn or their email. Back in the day, and 98% of B2Bs are still using them, you’d use Clearbit or ZoomInfo or Demandbase or Snitcher to get an idea of a company that had visited, like, hey, HP visited your contact page, or Microsoft visited this. The problem is Microsoft has 200,000 employees in different countries, many of them interns, assistants, non-decision makers, not super relevant. So it’s mostly misleading. Now with some of these tools we can capture and track down specific individuals.

The problem is, most B2Bs we work with set this up, sign up, and say, we need as much traffic as possible because we no longer need contact forms since we can see 20% of the traffic ourselves. So they end up collecting 3,000, 4,000, 5,000 leads a month and say, now we don’t know what to do, so we need to engage in email outreach and LinkedIn outreach. So the bottom line: what’s the problem, how to manage it, how to scale it, and how to actually solve it and make money through customer engagement. You built incredible content on your website, infused it with paid and social, so you get all that audience, they visit your website, and then you have a reason to have a conversation with them. Most B2Bs hand these leads to their B2B teams and start calling right away. This works in some spaces, I’m not super fond of it, but if you get 4,000 leads you don’t want to spend 100% of your time emailing, you want to first filter the ICPs, your ideal customer profile, and be personalized. So tools like Clay that we use do the AI-powered personalization, they can fetch based on a company, scrape LinkedIn, tell me what this company is about, what’s that role, and based on that role we sell this solution or product, based on that role we have these case studies. So you map out an entire engine of, if a marketing director from a company of 500 people comes in, then I’m going to pitch them a solution we built for a similarly sized company. So it’s completely automated and personalized.

I really hate spam, I get an absurd amount of spam myself. However, when I visit a website and I’m browsing their landing page and they reach out to me, well, I’ve obviously been doing research on them, I saw them somewhere, did some due diligence, so if they reach out I don’t consider that spam. They may have something else to offer, a way to enrich that data, a case I haven’t found, a discount code. So it’s a lot more natural to do it this way. If you don’t want to use Clay, a couple of other options are connector tools such as Zapier or Make, and an app extension such as SheetGPT. With SheetGPT you can go to Google Sheets and type a prompt in a cell and say, hey, use the previous cell and combine that and run a prompt and produce something else, and then drag and drop that for the entire sheet so it repeats the same process for all the other cells.

We’re getting closer to the end, so I’ll pace up. Sendspark is another one. You can record an intro video or a brief breakdown of your case and pull the same list from Clay or RB2B or Sidestep into Sendspark and say, hey, I want to send everyone this personalized video and just change the first five seconds, hey, first name, really happy to chat with company. So similarly to my creepy bot, you’ll have a full recording of yourself, so you’re the real human, but there are these two seconds where the AI fills the gaps to pronounce a new name and company name, and you can produce a thousand videos for an email list of a thousand people and auto-blast them, once you’ve just recorded a single video. So you can see how much closer this is getting to personalization while still providing the value.

The customer journey is pretty complicated, we have a lot of other aspects you can dive into yourself for customer acquisition. Specifically for paid ads, what we’ve been exploring more closely are tools like Magic X to run meta campaigns and scale. So we’re trying to connect the ICP persona builder to client research to ads to Midjourney creative generations to Magic X, just move the entire flow to AI. For customer onboarding, we use chatbots and virtual assistants. For customer support, we use different tools, Kustomer is a specific CRM providing AI agents, they used to be part of Meta, we worked with them for several years, really specializing in AI customer support.

To sum it up, AI isn’t flawless, but the thing is, experienced professionals scale 10x today. I’m going to quote something I saw on Seth Godin’s blog, and I’m going to rephrase it, Seth has a daily blog which is kind of high-quality. What he said today was roughly the following: give AI most of the work you do that AI can do, and use all the free time to learn and implement skills that AI cannot do. So smart people understand that AI is taking jobs that are mundane, copy-paste jobs, and what we need to do is empower AI so that we are 10x more productive for one-tenth of the time, and spend the time on being smarter, more engaging, conversion rate optimization, and providing that consultancy and expertise that’s really the value added on top of the collection of global knowledge that chatbots now have.

So that’s pretty much it. I’m sharing my resources if you want to know more, look me up, Mario P on Twitter and LinkedIn, my blog is mariopeshev.com, and I’m adding a link to my newsletter, which I’m posting weekly, go-to-market strategy tips. So thanks for your time, if you have any questions I’d be happy to be here. Shel, it’s in your court.

Host: Honestly, I’m going to pivot back to you because we got some good questions. Phenomenal presentation, thank you for that. Let’s see if we can get some of these up on the screen. So, how can AI help identify and prevent fraud in e-commerce transactions?

Mario Peshev: Yeah, there are several tools. First off, I use Stripe, and one of my companies is a Stripe Atlas company, even though I use Doola for all my US LLCs. The Atlas company has Stripe Radar, which incorporates a ton of AI and primarily a ton of machine learning to identify fraud. And now I see more alternative solutions coming up for identifying fraudulent transactions. For the most part, you can bet on different factors, in most cases it’s a matter of using an authority data source to feed that data with enough insights. The reason companies like Stripe or banks provide that is because you have a lot of credit data of success stories. Back in 2011 I was building a KNN algorithm for a bank for credit scores for credit loans, based on their data set of hundreds of thousands of people who had taken loans, so we were able to say, most people who are 21 years old, work a job making that much, and are single or married, 78% of them are returning their loans versus others. So based on these you can map machine learning, otherwise trying to do it yourself you can get a lot of false negatives, so definitely rely on a data set that’s trustworthy.

Host: Great answer. How can AI improve dynamic content generation for automated marketing campaigns?

Mario Peshev: Dynamic content generation. On the one hand, when I hear dynamic content generation, I hear programmatic SEO, and I also hear personalization in real time, so two things that we are doing here. On the one hand, programmatic SEO, for example we work with a directory for firearms and we enrich a lot of these pages that have very little information with regional data and seasonal data, pulling from different APIs and providing context, hey, what’s the temperature, what city they’re in, you can pull data from Wikipedia, but then for this category of business or type of reader in that city, come up with something new and unique and rephrase it for that persona and reader. So it’s about building on top of what we had five years ago and putting in more data points. And second, we use conversion rate optimization tools and A/B testing tools and real-time dynamic personalization engines to fetch data in real time, such as data from RB2B or Retention.com, and then below the fold or above the fold plug that in and use the city or the company and the person. Even now if you’re using Clearbit, even two years ago, if they detect that in their system you can say hi John, or hi, we love Microsoft, if they work at Microsoft. So this is evolving and getting more accessible and easier to plug into day-to-day business.

Host: Wonderful. We have a lot more questions but we’ve actually run out of time. I would recommend y’all go over to X and follow Mario. Mario, we actually have you next on stage as well, which is a very exciting session called the AI Startup Accelerator Game, and Brent Weaver will be joining us back. But would love to hear really quick, how has your experience been at Prepathon?

Mario Peshev: I really love the Prepathon. I’ve actually invited some clients and partners we have here because the conversations we have here I just send out to people and say, hey, these are the topics we’re working on today. These are the modern conversations of today. We’re talking about website optimization, AI-powered revenue and growth, AI and innovation. These are the conversations that everyone has to be discussing on a daily basis. If you’re not up to speed, surround yourself with people who talk about these things, because everyone cares about how your website gets better, faster, more powerful, more secure, more accessible, how to grow in that B2B recession space, and how to incorporate as much AI as possible for solopreneurs, entrepreneurs, small and midsize businesses and even higher. So everything relevant that you need, you can get it here at the Prepathon.

Host: I couldn’t agree more. I’m going to divert back, hopefully we have Brent about to join us. Thanks everyone.