Minter Dialogue with Michael Katz

Michael Katz’s Background

• Michael Katz describes himself as a technologist and a Baltimore native.
• He shares his passion for art, music, and innovation.
• Michael reflects on his journey from being an early adopter of new technologies to his current work.

AI and Technological Skepticism

• Michael discusses his initial scepticism towards AI and the hype surrounding it.
• He explains how his perspective shifted as he explored AI’s potential in his work.
• The conversation touches on the societal tendency to be cynical about new technologies.

Entrepreneurial Journey and Product Development

• Michael shares the story behind his new product, Flowsend AI.
• He talks about the challenges and motivations that led to its creation.
• The product aims to automate and enhance content creation for podcasters and other content creators.

Practical Applications of AI

• Michael and Minter discuss the practical uses of AI in business, particularly in content creation and marketing.
• Michael emphasises the importance of maintaining a brand’s unique voice while using AI tools.
• They explore how AI can help repurpose content and improve efficiency.

Challenges and Future Directions

• Michael outlines the challenges of running a startup, including funding, management, and client acquisition.
• He highlights the importance of prioritisation and learning from what tasks are left undone.
• The discussion includes the future potential of AI in creating personalised and high-quality content.

Flowsend AI: Features and Benefits

• Minter shares his positive experience using Flowsend AI for his podcast.
• Michael explains how the product works and its benefits for content creators.
• They discuss the importance of adapting AI tools to meet specific user needs.

Final Thoughts and Contact Information

• Michael offers a promo code (Minter 30) for listeners to try Flowsend AI.

Key Takeaways

• The importance of maintaining a unique brand voice while using AI.
• Practical applications of AI in content creation and marketing.
• The challenges and rewards of entrepreneurship and product development.
• The potential of AI to transform how we create and consume content.

Please send me your questions — as an audio file if you’d like — to nminterdial@gmail.com. Otherwise, below, you’ll find the show notes and, of course, you are invited to comment. If you liked the podcast, please take a moment to rate it here.

To connect with Michael Katz:

• Check out Flowsend.ai here & use the promo code Minter 30 for a 30% discount offer
• Find/follow Michael Katz on LinkedIn
• Find/follow Michael Katz on X (formerly Twitter)

Further resources for the Minter Dialogue podcast:

AI adoption | AI skepticism | Ai Technology | art and creation | Baltimore | brand voice | client acquisition | content repurposing | content transformation | event content | flowsend | free trial | Gilman School | innovation | LLMs | media companies | Michael Katz | music lover | personal voice | podcast automation | podcasting | prioritization | proprietary AI | sales technology | SEO keywords | startup challenges | technologist | transcription services | writing amplification

Transcription courtesy of Flowsend.ai, an AI full-service for podcasters

Minter Dial: Michael Katz, how great to have you on my show. I met you through our mutual friend, Sam Sethi. We have a mutual passion in the form of podcasting, and we’ve had several encounters on, let’s say, in the United States and in England, and always enjoy chatting with you. And so, I thought it’d be wholly appropriate now that I’ve gotten to use your new product, to have you on the show, talk about your career and your, your new baby. So, in your own words, who is Michael Katz?

Michael Katz: It’s a complicated question. I am not someone who is defined by my work. Okay. But it is something, obviously, I spend the majority of the day and night on, so of course it is. But I would say I am a Baltimore on. Certainly take a lot of pride in being from Maryland, and that is really at my core. But I am a technologist, I guess, by trade, purely as someone who grew up as the annoying friend who would download any app that was new and try to, without success, convince everyone I know to use it. I’ve always been interested in progress and novelty and innovation and those kinds of things. But really, to my core, I’m someone who’s just interested in art and creation. And I’d say my core, I’m someone who loves music and watching other people create interesting things.

Minter Dial: Well, that is glorious. Of course, the Baltimore connection is all the more personal because that’s where my sister and her whole family live. And on top of that, you went to Gilman, which is where my wonderful nephew goes. He’s Broderick. He’s finishing up there. My two nieces that went to RPCs. So, those are initials that you would understand, but not everybody who’s listening. These are private schools in northern Baltimore. And anyway, we also have a lovely appreciation of good music. Michael, and let’s start with your geekiness and how much AI has been a part of that geekiness that you were talking about.

Michael Katz: Yeah, well, you know, it’s funny, I’ve been through, like, a little journey on this front because, I don’t know, maybe like anybody else go, I go through certain, like, changes in my own life and preferences and how I feel about things. And I think for a little while, while I was someone who, as a youth, was, like, interested in the things that no one else was and wanted to use new things, I think I went through kind of a cynical phase in my twenties where I thought I didn’t have time for anything new and kind of went through that, like, jadedness of, yeah, this, this too shall pass. And you know, let me wait to be kind of in the middle, middle tier of adopters of any given thing. And that actually happened with AI, where I felt really skeptical about it and almost annoyed, I would say. And I think a lot of people are still going through that phase a little bit because any kind of hype cycle becomes very annoying, uh, objectively. And you’re just like, I want these people to shut up.

Minter Dial: Not for ready.

Michael Katz: Exactly. And you’re like. And you have a bad experience with it. And you’re kind of like, this is the dumbest thing ever. And, and let’s just be honest in modern society, not to make a philosophical point. It is much more fun to be very skeptical and to make fun of something than it is to be optimistic and excited. You kind of look smarter if you’re cynical or more skeptical. And maybe that was related to my own, like, proclivities. In that sense. I was just like, yeah, it’s more fun to just crap on something. I was really skeptical of it. And then we were building a totally different product, the team in the sales technology space. And, yeah, it turned out that some of the things we did, we decided to explore AI. Yeah, this could benefit from that. And long story short, I started to realize how powerful it could be. And for me, as someone who has always had a strong passion for writing and expressing my own creativity, I realized it was something that could amplify mine and give me more time to do that or program it to my preferences and allow me to do even more. And so, I’d say beginning of this year, we really decided to lean all in on that and to start an app where I had started a podcast from my previous business. And I, I wanted to go automate a lot of those components on the backend and spend even more time, almost in a way, to kind of tailor that. And so, I couldn’t find anything out there that I really loved, I thought was up to my standard. And so, I decided to build it myself. And that’s the kind of progression I’m on right now.

Minter Dial: Well, it is the classic case of have a problem to fix kind of entrepreneurship, and you can solve that because you know about it. And what’s interesting about your journey there is, for the majority of people who are listening, they’re typically in business that is either bereft of or looking at AI. And it does feel like the question of if it’s relevant is no longer relevant, it’s when and how much it’s relevant is really the more salient question. And to what extent you might actually even convert your product or your service into a. An AI machine of some sort.

Michael Katz: Yeah, I like. I fully support people who are sitting on the sidelines. I have no. If you’re solving a problem or doing something in your life that you enjoy already or that you think is going really well, there’s no. I’m not going to say you need to push yourself to do something that new. There’s plenty of work you go do in any given moment. But what I would say is, actually, I’ve heard this is maybe a French term, the appetite grows with the eating. And that’s kind of how I feel about this, which is like, you start to use it a little bit, and you start to realize, hmm, I wonder if we could do this? And that’s what we’ve seen people we use with our product of, like, okay, they started taking back some of their time first as a basic thing, and they’re kind of like, I wonder if I could also do that. And they see that, and, like, that’s pretty magical. Maybe I could do more. And so, that is. That is the challenge we all face ourselves with right now is almost like, yeah, be careful. You might get a little too obsessive if you start to really explore the different kinds of things and imagine, maybe this is something I don’t enjoy doing, or this is something I wish I could do even more intensely. Can I do it and try it out and see what happens?

Minter Dial: It brings up the notion of being cynical, as you said, being cynical, critiquing. The French also have a good saying for that, or at least her pastime. And you’re right, though there is a French expression, although the exact wording escapes me about that. The appetite makes the. No. Eating makes the appetite grow. It’s more along that line. But the idea being cynical looks easier. But then when you actually open the pandora’s box and you start saying, oh, my gosh, I it can do this. Huh? Wonder if it can do that. And then the appetite grows. And how do you curb your appetite when you realize how many more things you can do? And it’s endless possibilities.

Michael Katz: I find that challenge every day. And, I mean, look where we were at as, like, a business. And I’ve not always been passionate about podcasting, per se. I’m no mentor dial, not prolific in that sense. I started a podcast more opportunistically because I thought the human voice and having conversations is one of the best ways to communicate and to share your thought leadership or share your thinking and to interview other people and learn from them. And I thought podcast is a really interesting way to do that. It’s almost selfish in a way. By highlighting other people, you get to learn and you get to share their voice on your platform. Wow, that’s.

Minter Dial: You’re breaking my code secret. That is why I do this stuff.

Michael Katz: Yeah, it’s like I could ask someone to have a conversation and they’ll just be like, why? And if I’m like, hey, come on the podcast. They’re like, yeah, sure, when do we. And I’m like, wait, that is very bizarre. We have a very weird, like, conflicts around this thing. And I started, so I started that. Cause I said, well, I have this imposter syndrome. And I said, I don’t know if I really have anything to share about the sales space. Why don’t I interview other people, entrepreneurs I respect and sales leaders I respect. I was astounded by the quality of people I could get on. I said, what? You’re like a billionaire and you’re like, you’re more nervous than I am about recording this thing. That’s really crazy. Well, you have an ego, too. That’s really wild. And so, my goal was have these conversations and meet these people and learn from them, and then ideally turn it into what I would describe as a content machine. So, how do I take what’s there and transform that into other pieces of content? And that is where I thought AI could be really powerful because I was having a hard time doing that myself. And to your point, curbing my own desires to what else I wanted to do. And I want to create blog posts, and I want social posts around this, and I want short videos and I want all these different things. And I needed a tool for that. And where I think AI is really powerful, or maybe I’ll start ascent by saying where it’s not very powerful today, in my opinion, is creating quality, original thought and original content. It will get there, I’m sure. I don’t want to, and I’m going to be eating crow in 5-10 years, but I’m also just not that interested in that. Personally, I think humans are amazing, and I want us to share what we think in some kind of original way. The world’s a worse off place when we have less of that. And so, I was more interested in this kind of use case or this, this practice of taking original content and transforming that into, into other ways that people can consume that, or other ways of chopping that up so people could ingest that more easily. It’s hard to break through the noise. Right now, there’s so many different places that people are at any given time and channels, and people want to consume, but they need to do it in a way where it meets them, where they are. And so, that was what I was really interested in. And I think what we’ve done over time is be like, well, podcasts are not the only thing that people record naturally and have conversations on. What other kinds of informal speech and audio can we transform into these things people look for overall?

Minter Dial: All right, so, agreed. And let’s dig in on Flowsend.AI, your baby here. Tell us about its origin, when it came around, how you got up with the name.

Michael Katz: Yeah, I guess we were not feeling confident about the future of the sales software that we were building. We had some customers, and it was going fine, but I had my first child last August. Congratulations. Thank you. And that forced me to confront some really interesting things about myself and what I wanted to do and what I enjoyed doing and what motivated me to leave him. Right. I was like, I don’t want to ever leave him. The only things that I’ll leave him for are things that, like, motivate me more naturally. And so, I said, you know, I just, I think we’re running an uphill battle here, and I’d love to kind of sell this off or break this technology we already have and figure out what else we’re interested in. Or else I’d rather have a perfunctory job that I don’t particularly, you know, care about stopping at a certain time. You know, I work hard, but I’d rather just be at work for someone else and be told what to do a little bit. And so, we were exploring kind of our curiosities. And that area of the podcasting one was something that I’ve been really annoyed by because it’s something I enjoyed doing, but then I didn’t enjoy that part of it. And so, I asked our team, could we create something? And there was other software out there, other AI. What I found about it was it didn’t adhere to my standards of, like, I’d call it brand voice or personal voice. Um, it didn’t write things in the way that I wrote things. I had to train it all the time. It was a repetitive process, and I care about that. Maybe 90, 95% of people don’t care about that. They just want anything, and they can use our tool. I’m not going to stop them, but they might not convert to something from something else. But for those that do, whether it’s companies or individuals really care about that. We want it to be like, input your brand voice in a way, input what you want in a very customizable, personalizable way. And that’s what we would really want to focus on generating for you is taking your original content and conversations, turning them into outputs that really sound like you. Um, so you don’t have to go do a ton of editing and can still like, you know, not necessarily. If you want to, if you want to spend the same amount of time, great. Have high quality, um, in terms of the name, I don’t know, it was on a whim. I just thought like, what are some things that I want to, um, convey some notions. And I, we think like, people are interested in tools that get them to the outcome they want. So, maybe if they want to take a beach vacation, how can this tool help you take a beach vacation? You don’t really care about the things that it, you know, the technology itself, it’s like, that’s secondary. And I just thought, I don’t know, beach, flow, send flowing. And I thought, yeah, let’s, let’s start to build a more branding around. Like going to the beach, surfing, things that people enjoyed for better clarity.

Minter Dial: It is not Flowsend.AI, it is send.

Michael Katz: Well, now you’re inspiring me. Now you’re inspiring me. And also with the send, I was like, I want you to be able to work more effectively, but also be able to deploy this stuff in a way that is very natural. And so, it’s the outward sending of it all.

Minter Dial: We’re going to get into the product itself in the next moment. When you are talking about creating voice and repurposing content, I do absolutely agree with you that the future of AI in branding and marketing will be on creating a proprietary standout voice and that will require brands to understand what is their voice, which I would say is a dark mystery for the grand majority of everybody, as in who are you? Like you said at the very beginning, when I ask you who you are, it’s a complex question. And I think most brands have a rough idea, but don’t have a precise idea and therein have a very bland brand message or a very bland brand voice. And so, understanding what is your voice is probably one of the most salient outcomes you can have by looking at your product and what you’re doing.

Michael Katz: Well, I think it’s interesting that you said that. I think people shouldn’t put too much pressure on themselves to immediately have that voice. You look at some of the great artists of history and a lot of them started as, like, pastiches or copies of other ones. I was just looking at yesterday and thinking about Bruce Springsteen, who’s considered this very unique American voice and rocker, basically a Van Morrison copy. I mean, his early songs sound just like Van Morrison songs he’s copying. Right. And Billy Joel takes different tones in his music, too. He’s a New York. No, he’s, like, written songs that are literally in the exact style of the Beatles. And that’s, like, you’re kind of playing a part for a little bit. And I think that’s actually what people should do is look at someone they like and copy them and imitate them and play the part a little bit and try on different voices to see what they’re interested in. And through that process, you kind of find yourself unless you have this crazy vision already. And so, some people on our platform, and you can to kind of get into the little nitty gritty, you can put in examples of what you like or what you wanted to imitate. I’m like, you know, if you have someone you like already, but you’re like, imitate them, that’s okay. Like, just start with something you enjoy and imitate elements of that. You’ll find something that, you know, you’ll find your own voice. So, I do agree that is, the future is, like, being able to stand out, even if that is something inefficient and weird. I mean, I’m planning on starting, like, a live webinar and podcast series at night. And that’s terrible for the algorithm. It’s terrible. It’d be the dumbest thing ever. But I’m like, but I enjoy it. And I’m a night owl, so, like, that, that appeals to me. Just, like, being a little bit different, doing something that’s abnormal.

Minter Dial: Well, when you are trying to please everybody or do everything for everyone, you’ll get nowhere very fast. And I totally agree with you about this idea of trying out new things as much like, you know, asking a teenager, what do you want to be in life? Well, shit, I don’t know. I haven’t tried anything. You got to do stuff, break your teeth, figure out you like, you don’t like, and then you come around getting yourself. My point is more about the bigger companies that have been around. And in a world of trying to be precautious and safe in everything we do, we end up eliminating anything that has any edges on it, and we end up with this very bland voice. And. And I think that people aren’t really conscious of the idea of having a separate, maybe even niche voice that is daring to stand out, not on everything, because that’s like everything, you know, trying to just please everyone, but on something and make you stand out. And then, you know, maybe it’s, you always want to have in every headline, darn. Or shoulder or something. I don’t want to refer to head and shoulders, but you want some kind of thing that pops out. All right, well then everything you’re going to have is around this notion of pink. Well, what does that mean in terms of your voice and how does it, but anyway, going to your product, Michael: Flowsend.AI? I want to just take a second to speak about it in my own voice, so to speak, which is that I’ve been using it now for maybe three or four months, and I use it for all my podcasts. And it’s amazing how it can take my conversation, do the transcription, and then spits out ten or twelve customized pieces of content that I’m now using. I’m not going to say blindly, because I think that would be a mistake in general, but very quickly able to copy paste and push in into my social networks in repurposed fashion based off of my voice. And I find it fascinating. And so, congratulations to you and the team for doing that. My feeling is now, how are you approaching the future of this product and specifically looking at how to stand up and get a good business model?

Michael Katz: Yeah, it’s everyday changes. And I know that sounds cliche, but it is very real where we started with, as you already mentioned, scratching your own itch. Now we have this podcast and we want to create content around it. And then I, like, I think anybody should do, and I was used to building these like, enterprise products that needed a lot of people using them to even like in a company to even make them useful. And this is a product that suddenly I could give to people that was one person can use it and do one thing and immediately see what’s going on here. And that’s very exciting, or it’s really, it’s hard, but it’s exciting. And so, I gave it to a bunch of people, acquaintances, friends, marketers. I respected podcasters originally, and I said, yeah, just go use this thing and report back. Right. We’ll watch you and see what you do. And you saw that I started to see some people doing funny things with it and putting in YouTube videos and whether it was for their company that were like just product demo videos. And I said, that’s weird. How’d you put that in there? And they’re generating these custom descriptions for better YouTube SEO and generating blogs from them. And I said, that’s funny, right? And more and more I started to realize, well, I guess on a generic level, this can take any kind of audio video content and produce other kinds of other kinds of content. And what we really are focused on today is written content. So, the platform, I guess, is kind of like, I describe it as two functions. It’s number one, a really good analyst, someone who can operate at 1000 times the speed of a normal human to come back and tell you things about your recording. And then number two, it’s a copywriter. And I think we have plenty of room to run on the function of both of those. But how do you do everything you would need from those types of people with your recordings? And so, we’re taking things like webinars and events. I worked at a company that spent many, many millions of dollars on an event every year with amazing speakers talking about what they were doing with the platform and their own thought leadership in the industry and how much they love what they were doing. Nothing happened with that content. And I thought that’s really sad. You have these understaffed teams that need good content and it’s all locked up in these videos and you’re not doing anything with it today. That’s a use case. I’m really passionate about taking events. Anyone listening to this has an event coming up that they are attending or that they are hosting with a bunch of talks or working with people on that and co-branding assets and saying, hey, look, go check out what we did and feel like you didn’t miss a thing, even if it’s just the summaries and key takeaways for these people. So, that’s something we’re interested in. We’re signing a few deals with media companies to automate work for a lot of their people and standardize and help their people try to spread their own reach and audience through promoting their content better. And then some of the stuff we’re doing is really simple. And I just reached out to you about this last week, which is we have, I’d say, industry standard, industry leading transcription for AI. We’ve discovered some techniques that’ll get us to really, really special and we’ve had people talking to us about that who pay people today to transcribe their work and taking these opportunities as they come, because that’s a lot, you know, it’s a lot easier to sell someone on something they’re already doing and say, we can do that just as good, if not better, cheaper. So, really taking things as they come.

Minter Dial: It strikes me, Michael, that this notion of text is fundamental, because that’s how the Internet works. Whether it’s ones or 0’s or texts that are tagged into videos, this is how it’s searched, this is how it’s the primary method. And so, being able to move video with sound or an mp3 into text, that’s become searchable and then you can do all the modifications behind it is a fundamental piece. So, I have a lot of friends who are doing speeches, and of course, there’s so much people talking about how videos is everything, yet is there a need to get into the text, and then you need to get the video to be seen, because like you said, the beginning, there’s a lot of noise. So, I might have got a great video, but nobody’s seeing it. So, how do I get people’s attention? Well, how can I quickly break this down into three punchy words or three punchy sentences that I could put on social media one way or another, or send an email blast or whatever, so that people will do it? And for having been doing this for now 15 years, you obviously get into habits. You think you’ve got an idea of how to get engaging content. And the benefit for me, by using Flowsend, has been able to spark a whole variety of different ways of approaching headlines or engaging texts And so, on. So, I think it’s very exciting, and I certainly am going to promote this to all of my friends to go check it out now for people who are listening. Few people have not heard about LLMs or large language models, and certainly about artificial intelligence. There are today, apparently 100 new AI initiatives happen every day. If you’re in a company or in any business, and you’re looking at saying, I want to use these LLMs, what would be confusing for me is how to figure out the path to making a unique proprietary AI. That’s not going to cost you an arm and a leg, because we are no googles, we are no apples.

Michael Katz: Yeah, well, that is, I mean, I heard some stat recently that these big consulting firms Accenture all those. It’s like it’s becoming half of their business. McKinsey, et cetera, is consulting on AI. Now, I have respect for many of the people that work at those organizations, uh, and friends. I think some of it is probably like putting out a deck and saying, here’s what you should be doing, and probably not going to result in anything too soon, because that’s the pace of people and corporations and the way they move typically. And, um, so a lot of people are thinking about that. I talked to a news organization recently who was saying, oh, we’re paying someone to build a proprietary AI model on archive. Like, I don’t, you know, I just have my, again, the skeptic in me arises again saying, yeah, I’m curious what’s going to happen there? The things are moving so fast that, like, you might actually be behind if you’re already building something in a weird way for certain things as opposed to waiting. But, yeah, so that’s, that’s what’s complicated. And so, like, I think where people need to be focused on and think about using is like, tools that respect your voice or your content, and that can take in a bunch of your, um, preferences and, and, you know, your files and your content and your text, whatever it might be, and, um, solve specific use cases for you, because I think that’s where we’ll approach that. We always have approached technology that way. We don’t, like, care about the innovation itself. We don’t care about what chip it’s using. We care about what it does for us. I mean, that’s how we use a computer. It’s like, what can it do for me? I don’t care about what’s inside. Most, 99% of people don’t care about what’s inside of it. And so, that’s how I talk to people about it, is being like, well, what are you doing today in general that you don’t like or that you’re spending a lot of time on? And can that help you? I mean, the most basic use case, in my opinion, for AI is as a thought partner. I think that’s kind of the first level, which is like, okay, the things that back in the day, I would have said, what is this? Or, how do I do about this? And I go to a library, and then eventually you went to Google, and you’re like, google, what is this? And you got kind of some maybe fine, crappy results, maybe just good enough to go to Wikipedia and find what you need. And that’s the first thing people do here, is that now what I use it for on a day-to-day basis is we have a small company, and I can’t always talk to someone about something. And I will genuinely detail my thoughts to the level that I would to someone else I’m talking to as an advisor, and say, hey, look, you know, act as this kind of, for me, I am thinking about the trade-offs between these two or three things. Here’s what I’m thinking about, and I structure my thoughts. Sometimes I’ll talk to it and I look to understand what it thinks, and I can disagree or agree, but I think humans are pretty bad at taking their own advice or seeing the obvious answers in front of their face that kind of need someone else to tell to them. We’re very imperfect in that way. It’s why even though I think they’re fantastic, therapists will always have a job, sometimes you just need someone to tell you what to do. That, you know, is obvious. And so, um, that is the way that I think people should deploy them at first, is just saying, hey, use some of these generic tools and, and talk to someone like you would the smartest child you’ve ever met, who somehow knows everything but needs a little bit of specific instruction. Um, and then in the future, of course, we’ll have them kind of take over and know everything about our companies and deploy them, you know, far and wide of like these generic tools. But I think we’re, we’re a long way away from some kind of generic tool. Solving things for you. You need, like, this is my problem. Is there a solution for this right now?

Minter Dial: And this is what I certainly appreciate in flow Sen is that you absolutely have been adapting it to my needs as far as podcast is concerned. And going back to this idea of problem solving we need to lean into, if we’re in an organization, what are the problems we want to solve for our customers, and how can AI do it, but not do it solely? It’s going to have to be human plus AI. And one of the so a use case that I had, and thanks to your inspiration was I was doing a speech and I recorded the speech. I then downloaded it in a flow send, which then gave to me a recap key bullet points, things to remember the key keywords for an SEO purpose. So, all this stuff came out of an mp3, which allowed me to repurpose it. And when you are doing it that way, you’re getting more in the weeds, being more specific about how you do it, as opposed to sort of just using one of the generic LLMs out there. So, the question remains, which is, how do you concoct what cocktail of LLMs? Does it mean you have to know and improve on the voice to text technology that’s existing by itself, or does that mean you throw together two different services? I just have no idea how to navigate that type of path when you’re trying to deal with figuring out the solution to your problem.

Michael Katz: Yeah, well, we. Yeah, like, I think it’s a good progression with anything. When we see someone using chat GPT to do what we do, that’s fine by me. Actually, that was my progression, so great. That’s fantastic. People need to believe they have a problem and try to solve it before they start getting better solutions for it. You can’t convince someone just to do something. They need to know that they have a problem first and try different things. What we do is we are using multiple LLMs and we’re also constantly testing which ones are better for which tasks, for which type of writing. We want to abstract away that complexity from the user of what do they need to choose and where do they need to turn to and how do we transcribe things better for them and then how do we create better content for them and what are the types of content they should be creating? And we’re trying to create a platform that is at once simple enough for anyone, like a philistine who doesn’t use any technology to use it, and just say, I can get value out of this thing. And for the most complex people who are like, I want to really customize this thing and rev it up because I know exactly what I want and I want more outputs than what we give you. And you can plug those in and that’s my new template. And so, there’s the like choosing your LLM and then there’s the understanding how to talk to it, which that’s an art. That is an art all its own. And people talk a lot about, oh, you’re going to hire these prompting specialists, prompt engineers, and that’s a real thing. There’s software that will emerge to help with that, but how do you know to get, there’s some things that help you get 1% better results, some things that help you get 10% better. There’s some things that’ll take away some of the AI’s worst ticks and word tendencies, and that’s part of our challenges. How do you make something seem really simple that wasn’t actually that simple on the backend? And for those who can’t tell the difference, fine, so be it. Go use whatever you want to use. For those that can tell the difference, we want them to be impressed. I remember hearing that Kayak would artificially make your loading pages longer when you got a flight because they wanted to show you this was hard, even though they already had loaded the results. And it’s a little bit like that where I’m tempted sometimes to be like, should we make this seem harder? Because then maybe people will appreciate us more.

Minter Dial: But, yeah, I’ve been exploring that with the idea of authenticity. And the first idea was, make it slower, because if you use a site like cookie AI, which responds as fast as you’ve hit the enter button, you’re like, well, that’s like, obviously isn’t a human. And then you sort of have a moment with that. Well, how do you make it look human? And one of the ideas is what it was banded around was, well, make it have spelling mistakes, because that’s really human. We are imperfect. Well, it turns out that when you make spelling mistakes and this is a message for all human beings, it corrodes your trustworthiness. So, it behoves us to actually be more anal about our spelling, which not everybody is doing these days. But I finish on just two more questions. One of this is, you talked about figuring out improvement. Is it 1%, 10%? And I find that there’s so much artistry to that, which is the right transcription. I mean, that’s word to word, but the right interpretation. They made chat GPT did a better job than Gemini on or whatever, on interpreting into one paragraph. And how do you debate the qualities of that? And then the other piece, which is, how do you remove the AI ness of it to make it sound more in my voice? Which is your first point?

Michael Katz: Yeah, well, we. I mean, first and foremost, we have always had to provide it with examples of what we think good looks like, and that has perspective. And, you know, people in software talk about this notion of, like, is your, is your technology opinionated or unopinionated? And there’s a lot of, there’s a lot of things that go into that, of like, you know, is it something that the user can just do whatever the heck they want, or do you guide them to do something very specific? And, you know, does it have these perspectives embedded in it? And this is an example of one of those, which is like, this is AI with some of our perspectives, and we need to keep enhancing that perspective. But, yeah, we, you know, we could generate really titles for your podcast that are really clickbaity. So, you know, these are the best for virality. But I think a lot of people would look at those and say, I don’t like that. That doesn’t sound too much like me. And so, we have to toe that line of, like, what is good? And then allow you to customize a little bit around what do you like? That’s good. And we’ll continue to think about ways of, like, making that easy for people. Um, but, yeah, absolutely. It is like, that is a huge, a huge debate that we have internally. Uh, how do you make it not sound like AI? How do you give it high quality? How do you make this thing natural? There are so many factors that go into that in terms of what we do on the backend and training it and telling it different things of like, don’t do that, do this.

Minter Dial: In “The science of storytelling,” a book by Will Storr, he talks about the complexity of the reasons that we are and who we are. And it’s really never a perfect ending or a perfect story. A last question for you. Michael, regards this notion well. Running a startup, there’s always going to be challenges of funding and management, recruiting employees and recruiting clients. Which of those, of those four is the hardest for you?

Michael Katz: List those four again.

Minter Dial: All right, so funding, management. So, dealing with the priorities and resources, recruiting good employees and getting new clients.

Michael Katz: Well, I think number four is the number one priority. That should be the number one priority of any startup, right, of getting clients. And that is like the end goal of anything. So, I don’t want to not say that’s first, but I do think that, like, prioritization is probably the first one for me. And those are intertwined because I am someone who gets excited by ideas and gets, you know, excited to do things and is impatient. And I have to curb that impatience on a daily basis of being like, what is manageable for me and something I work on a lot as a human being of. I get frustrated by what’s not done at the end of the day, as opposed to being grateful for what is done. And that is downstream from prioritization. And I think what I’m starting to realize over time, and this is a product of age and work, is prioritization appears naturally in what you do, which is what didn’t you do. And if that isn’t, like, you know, in line with what you think should have been done first, well, then you learn a lot about what you actually valued and what was actually important. And so, that probably is my number one of, like, prioritization. And then, you know, sometimes I think about how to unblock other people on the team, and I’m like, well, I just spent all day unblocking other people and I got none of, quote unquote, my work done. And maybe that’s okay. Maybe like, that is part of the job of management. And the whole conceit of it is like, it’s more important to get everyone else and to give them the tasks to do. And what you can do on the back end is just gravy at that point.

Minter Dial: Allow everybody get on with their job. Michael, been great having you on. Love your energy, and very much enjoying flowsend.ai, which I highly recommend everybody try out. What is the process for people becoming clients? Do you have a freemium model? Tell us a little bit about that, and then, well, sign off.

Michael Katz: Yeah, sign up. We do have a free trial, so you can sign up. I would say that we will have a promo code in the show notes here. Let’s just call it for now. We’ll call it Minter 30. And so, that minter 30 will give a discount for folks who do it through this podcast here, because some people say we’re too cheap right now. It’s hard. We want as many people as possible right now because it makes me more excited to serve. So, yeah, it’s flowsend.ai.

Minter Dial: And how could people get in touch with you or follow what you’re up to?

Michael Katz: I’m on LinkedIn. Michael N as in November, Katz. Twitter is Miguel Katz. My father’s Chilean, and Michael Katz is like the Jewish version of John Smith, so that was easier. That’s what he got. So, yeah.

Minter Dial: Muchas Gracias. Yes, Michael. It’s been a pleasure. We’ll stay in touch, my man.

Michael Katz: Yes, thank you.