Episode 82 - From Writer to Developer: Vibe Coding and Building Your Own AI Tools with Joseph Ucuzoglu

This week on Brave New Bookshelf, we’re diving into an exciting conversation about how the line between “writer” and “software developer” is completely blurring. In episode 82, hosts Steph Pajonas and Danica Favorite sat down with Joseph Ucuzoglu, a developer, author, and one of the co-founders of the Future Fiction Academy (FFA).

Known as the “father of Rexy” (one of the earliest AI sequencing tools for authors), Joe joined the show to discuss “vibe coding,” building your own custom writing tools, and how next-generation AI models are putting the power of software creation directly into the hands of creatives.

Meet Joseph Ucuzoglu, Developer and FFA Co-Founder

Joseph Ucuzoglu is the technical mastermind behind many of the tools used by the Future Fiction Academy community. With over 20 years of experience in Information Technology—including a long tenure as a Salesforce consultant—Joe first dipped his toes into the AI writing world around the release of GPT-3.

While trying to write his own book, Joe realized he wanted a better way to manage his prompts. He built a custom Chrome extension to organize his variables and templates, which caught the attention of FFA founder Elizabeth Ann West. This meeting of minds led to Joe building Rexy, a groundbreaking tool that allowed authors to automate prompt sequences and write back to Notion databases.

Today, Joe is a core partner at FFA, where he heads up the “Ink & Code” curriculum, teaching authors how to build their own software solutions. He has also successfully completed his first book draft, embracing the excellent advice to “never make your first book your most precious project” so you can freely experiment and learn the tools.

The Evolution of AI Writing and the Rise of “Vibe Coding”

The episode kicked off with a look back at how quickly the AI landscape has evolved. Just a couple of years ago, authors were struggling to generate coherent paragraphs, relying on tools like Rexy to stitch text together. Today, advanced models like Claude can generate massive, highly coherent outputs in a single turn.

But the real disruption is happening in software development. Joe highlighted the concept of “vibe coding”—a term used to describe building software simply by describing what you want in plain, natural language.

“We’ve gotten over the hump,” Joe explained. “If you are a person who doesn’t even know how to code, but you want to create a little app to simulate note cards on a corkboard for your writing, that is now within your capabilities.”

The Ultimate App-Building Pipeline for Non-Coders

For authors who want to stop relying on third-party software and start building their own custom tools, Joe shared his personal, highly effective development pipeline. Best of all, it requires zero prior coding knowledge:

The Brains (Claude Desktop or ChatGPT Codex): Use these tools to describe the application you want to build. They can read and write local files on your computer and generate the necessary code.
The Workspace (Cursor): Cursor is an IDE (Integrated Development Environment) that Joe highly recommends. It makes it easy to review the code generated by the AI and connect your project to GitHub.
The Storage (GitHub): A cloud repository that stores your code and tracks different versions of your project.
The Server (Railway): A service that connects to your GitHub repository. Whenever you update your code, Railway automatically redeploys it and provides you with a live, shareable URL.

By telling Claude or Codex to write the configuration files for Railway, the entire deployment process becomes a simple “one-shot” execution.

Visualizing Stories with Next-Gen Image Models

The conversation also touched on the massive leap forward in AI image generation. Joe and Danica discussed the release of ChatGPT’s Image 2 and Gemini’s Nano Banana model, noting how incredibly consistent they have become.

Joe shared an experiment where he used a chapter of a book to generate a multi-page graphic novel. Unlike older image generators, the characters and environments remained completely consistent across panels—even down to the placement of bushes outside a character’s house when they returned home at night.

While authors may still choose to hire professional cover designers (as Danica did for her latest series to save time), these highly consistent image models are game-changers for creating mood boards, characterizing protagonists, and mocking up sequel covers to show designers exactly what you want.

Empowering Authors to Control the Prompts

A core philosophy at the Future Fiction Academy is that creators should have ultimate control over their prompts and models. Joe explained that when you use commercial writing platforms, you are locked into their pre-written prompts. By learning to build simple, custom tools, authors can tailor the AI’s behavior to their exact genre and workflow.

Joe shared a sneak peek into his current side project: a visual notecard app built entirely using Claude. The app allows him to drag and drop plot points on a canvas and have an LLM generate scenes directly from those cards.

“I am terrible at front-end user interfaces,” Joe admitted. “But it doesn’t matter today. I just told Claude what I wanted, iterated through standard language, and it built a beautiful, working browser app.”

Favorite Tools & Recommendations

Joe and the hosts highlighted several essential tools for authors looking to explore productivity, image generation, and coding:

Claude Desktop: Joe’s top recommendation for general productivity, writing, and coding tasks.
ChatGPT (with Codex and Image 2): Excellent for coding assistance and incredibly consistent image generation.
Gemini (with Nano Banana): Another powerful option for highly detailed and consistent visual generation.
Cursor: A user-friendly IDE that integrates perfectly with AI to make reviewing and managing code simple.
Railway: The go-to hosting service for quickly deploying custom web apps.

Key Takeaways from This Episode

Don’t make your first AI project your magnum opus. Use a “non-precious” genre or story to experiment, learn the tools, and get comfortable with the workflow.
Anyone can be a builder. With “vibe coding,” you no longer need to know programming languages like Python or JavaScript to build custom writing dashboards or planning tools.
Establish a solid pipeline. By connecting Claude, Cursor, GitHub, and Railway, you can go from an idea to a live, working web application with minimal effort.
Use AI to bridge the communication gap with human professionals. Use consistent image models to create mockups for cover designers or editors to eliminate ambiguity.
Let the AI make design decisions. Don’t be afraid to ask the LLM to rank design ideas; often, its architectural decisions are highly sophisticated.

Resources Mentioned

Here are the links and resources discussed in this episode:

Transcript

Speaker: [00:00:00] Welcome to Brave New Bookshelf, a podcast that explores the fascinating intersection of AI and authorship. Join hosts Steph Pajonas and Danica Favorite as they dive into thought-provoking discussions, debunk myths, and highlight the transformative role of AI in the publishing industry.

Steph Pajonas: Hello, everyone, and welcome back to the Brave New Bookshelf. I’m one of your co-hosts, Steph Pajonas, CTO, CMO, I don’t really know anymore, of Future Fiction Academy and Future Fiction Press, where we’re teaching authors how to use AI in any part of their process, and we’re publishing AI-forward books on the press.

I have … I wear many hats at the Future Fiction Academy. I do lots of stuff, so it’s hard to understand what my title is most days, but that’s fine. I do a lot, and I enjoy what I’m doing. We are probably by the time this goes live, we’ll be deep into our new curriculum at the FFA, and we’re teaching people craft, we’re teaching people how to manage their business, we’re teaching them [00:01:00] marketing, we’re teaching them code if they wanna learn how to, make their own stuff as well.

So it’s a full education at the FFA. We definitely try to teach everybody what they need to know in order to be a, an author in this new paradigm that we’re living in where AI can do pretty much anything in your business, and that means that you can level up and be even more of be a bigger author or even a publisher, a publish, publish your own books, publish other people’s books.

Do whatever it is you wanna do in the publishing space, and that’s really cool. And that’s what I like doing at the FFA, and I also like writing my own books, of course, as well, and I’m deep into writing a book with Claude right now. I just told him this morning, I’m like, “You have to stop doing it in this style.”

It doesn’t always listen, right? So we have to be the ones that are driving the train at all times, being the human in the loop, because they just … Sometimes they just don’t listen. I keep telling him “Don’t write like that,” and yet [00:02:00] he is still writing like that every single time. I think that’s the pain of just using AI right around now.

Anyway, so that’s enough of me rambling. I’m going to hand off to my lovely co-host, as always, Danica Favorite. How are you doing today?

Danica Favorite: I’m good. I’m good. Yeah. Feeling those Claude-not-following-instructions pains at the moment. I had an amazing co-work automation, and it broke. And I did figure out why it broke and where it broke.

But oh my goodness, I do think that sometimes it is like chasing squirrels at a rave, but this is what we signed up for, right? So for those of you who don’t know me, I’m Danica Favorite. I’m the community manager at Publish Drive, where we help authors on every stage of their publishing journey, from getting their manuscripts formatted to creating metadata and AI book covers, then getting your books distributed to the widest audience possible.

And finally, once your books start selling, we can help [00:03:00] you split your royalties and also do some in-store promos. And, like definitely in the thick of things. I think by the time this episode launches, we will be into June, and we’ll be enjoying our summers, and I may even be meeting with the team as we continually wrestle with this whole AI question, and how do we help authors using AI get their books published to the widest variety of stores and maintain our relationships with retailers, respecting retailers, and really trying to figure out what the AI slop is.

If you were listening to our last episode where we talked about this idea of the AI slop, and then I already forgot the term that we coined, but, this idea of the AI really not just assisting you, but empowering you. There we go, AI empowering you. Because there is a difference in the quality, and we want those quality books to [00:04:00] come forward and not be drowning in the slop.

And so it, it’s something that we as a team are constantly discussing and really trying to figure out the best way to honor people and their processes and to get those quality books out there, again, to as many people as possible. And so yeah, I mentioned I was messing with my Claude automation, which was so frustrating, but it’s fixed now, or it should be.

We’ll find out. It’s supposed to run here in an hour. And so we’ll see. But I’m actually, speaking of summer, planning my summer trips using AI. So hopefully please let this be over by the time our episode launches. But with all of the stuff happening in Iran, and Europe is talking about running out of jet fuel.

And so I had to change what I wanted to do this summer. So I go every summer to Budapest for work. It’s fantastic. [00:05:00] I love it. We have our big team meeting then. We talk more in depth about the things I was just mentioning. And then I like to add on a little trip. Since I’m already in Europe, I pay for myself to go on some fun vacation.

And because of the airfare situation and all these regional carriers that I usually use may not be operating, I had to flip and say, “What about… What can I do instead? Here’s what I want to do.” And so I went to ChatGPT, and I said, “Look, here’s what I wanna do. Given that I can only do the train, or I’m choosing to only do the train, what do you recommend?

What do you suggest?” And it was amazing because it gave me everything I needed. And it gave me links to sites. It gave me links to different things, and I have some really particular things I wanna do. And I said, “Okay, so what if we did this, and we mapped out this?” And it did it. And so I’m really excited because still have some things to book, but if all [00:06:00] goes well, it’s gonna be a really fun summer.

And I like to share these things because it’s so funny the things we don’t think about using AI for, and what would’ve taken me hours of planning, I just literally sat on my phone one day, “Burp burp.” Oh, cool. Now I have my whole travel plan. So, um, always fun and always interesting to see what AI can do for you.

And so don’t have a good segue to today’s guest. But Joe is someone who has worked with the FFA from day one, and is somebody who’s behind the scenes and doing all the programming. And when Elizabeth Ann West says, “Hey, I wanna do this cool thing,” she calls Joe up and says, “Joe, make this thing.”

And Joe says, ” “Okay.” Uh, I know there’s a little more nuance to that, but that’s usually how the story gets relayed to me. But he really is just a really interesting person and is doing so many cool [00:07:00] things with AI, and I was telling Steph let’s see if we can get him on,” because it would be really interesting to hear what he’s doing and some of the interesting things that happen from a programmer’s side, I talk about how I’m not the programmer. I just go “Boop, boop, it,” uh, “Make this happen,” and it happens. But he’s the man behind the magic. Welcome, Joe. Joe, please tell us about yourself and how you got involved into this crazy AI world.

Joseph Ucuzoglu: Thanks for having me on. So I’ve been in the information technology field for a little over 20-plus years.

Probably the last eight or so I’ve been working as a Salesforce consultant, building out, people’s Salesforce instances, and, custom code and attaching them to other programs. And I wanted to write a book, and this was around the time that Sudowrite was, starting to come into fruition.

This is before, Turbo 3.5 came out, probably around the ChatGPT, the 3 [00:08:00] days. And I remember using that, and I was part of the community over at Sudowrite, and I decided to write a little extension tool in the Chrome browser that kinda keeps track of all your prompts and allows you to kinda enter some information for the variables, and then it would throw it into the template.

And then you could take that, kinda copy and paste it into Sudowrite and use that. And so one of the people on Sudowrite who knew Elizabeth decided to talk to Elizabeth and said, “Hey, you should meet this guy named Joe. He created this cool tool.” And so he set up a meeting and I got in there with Elizabeth and then, the first thing she said is, looked at me sternly and I think she was worried that I was gonna try to steal people away from Sudowrite which I told her no, that’s not it.

This is a tool for the community, the same thing that, I do for FFA.” And so you know, we talked a little bit about that and then she gave me some prompting tips and a few months passed and then she started her [00:09:00] own school, the FFA. She started talking about it. She’s gonna do the latest and greatest in prompting, and I really wanted to do this, but I kind of, when I write I just do it a- as a side thing as more of a hobby. Hadn’t published a book. So I was thinking like, “Okay, this is something that would be interesting to do but I can’t justify the cost of it as somebody who is just, plays with writing, doesn’t really do it seriously.”

But I wanted to, so I asked her, I said, “Hey, I’ll make some tools for the community if you let me join for free.” She said, “Okay, you’re on it. You’re in.” And so from there I built out a tool called Rexy which then got adopted and then, um, you know, we had kind of a, a deal going on where I was making tools for the FFA.

Eventually became one of the founders of the school, and I’ve just been, building tools for the school ever since, and also working on some of the educational programming now, as we’re revamping [00:10:00] that.

Steph Pajonas: Yeah, Joseph is like the father of Rexy. It was so much fun to, at the time, we didn’t really have any sequencing tools out there, right?

We had nothing. We knew that we wanted to prompt in sequences, like prompt to get a response, tack that on to another prompt, go back, and then have that kind of automated because what we were doing at the time, we understood the prompting and the kind of responses we could get from that. It’s a lot different now that we have thinking models and whatnot, but back then it was like prompt response, prompt response in a circular like sequence thing.

So we had Joseph build us Rexy, and it was so interesting because we kludged together a bunch of things to get that thing to work, right? It was like Rexy worked on a website, but it hooked into your Notion to grab, uh, grab files from there, and it had access to databases, and it would [00:11:00] write back to Notion when you were done.

So this is how we got really into Notion, and now I’m a huge Notion fan, and I’ve mentioned it in previous podcasts many times, so I won’t go into that too much. But Joseph was the person who got us Rexy and over time, Joseph became one of the founders and became one of our partners and now he like heads up lots of the Ink and Code stuff, the coding stuff that we do on Future Fiction Academy, and then has been teaching people as well, which is great.

I love that. Joe has even managed to write a book now, haven’t you, Joe?

Joseph Ucuzoglu: I did. It’s still, I have not published it yet, haven’t hit the button. So I did put something out. I was given advice, don’t write your, anything that you’re precious about as your first attempt. So I’ve been…

and it’s been really nice. That was great advice because I’m using the tools and using AI to help with a lot of the drafting and editing, and I’ve been really just enjoying that [00:12:00] process a lot. You know, I, I feel like it’s really amazing how different the landscape has got from when I first started to where it is now and I can already see where it feels like it’s gonna go next, right?

We started out with the tools, like, where you could formulate maybe a few sentences, a paragraph, and then Rexy was there to try to see if we could stitch together a bunch of paragraphs so we could get maybe a whole chapter out. And and then we created our sequencing tools, using the different methods that we have right now.

But, now we’re starting to see these more powerful models like the latest Claude Opus 4.7, when it listens to you, and, it can produce some really large output that is seemingly coherent, which is really nice. And so I feel like we’re not that far off from the point where we’re gonna be able to tell the LLMs, with instructions, this is essentially what I’m looking for, and it- I feel like it’s gonna be able to handle a lot of that [00:13:00] writing.

You always want the human obviously to be there so it can execute your vision, but I just, I find it fascinating how powerful the tools have gotten and how interesting that is. And I feel like almost we’re getting to a point where some of these other things that you wanna do, like editing and so forth, are starting to now become something that the LLMs are capable of doing.

And with these new agentic tools and these harnesses that are allowing you to run these tools, it’s getting to the point where you’re gonna be able to almost, tell this thing, “Hey, go out there and make me money,” and then you can go and write your book on the side.

Danica Favorite: I think that would be fantastic.

I look forward to that day personally. Um, but yeah, I, I love what you were saying about how that first book attempt isn’t a precious book. It’s just something to learn on. And I hadn’t heard that advice before, but it’s something that I’m [00:14:00] doing, again, with my own AI writing experiments, is I’m using genres that I’ve never written in, that I don’t really, I hadn’t really seen as, ooh, I should go for this or whatever.

But I have friends who write in the genre, and I read a lot of it. So I was like, oh, let me just try here. Because you’re not as emotionally attached to it. It, it isn’t like, oh, it’s gotta be perfect. It’s, oh, it’s my magnum opus. It’s, no, it’s just a book. And so I really like that you did that.

So are you gonna get more books out there now that you’ve gotten that first book conquered?

Joseph Ucuzoglu: I think it’s interesting because i- if you think about at the very beginning, the whole idea of joining the FFA was this idea that I really wanna learn the crafting of the prompts so that it can help me be able to actually, produce books.

And then I found myself now so entrenched in the world of helping authors with tools and methods to be able to write their books that I’ve actually really slowed down the progress on mine, which I don’t think is a bad [00:15:00] thing ’cause I actually really enjoy coding. It’s like you’re taking two passions and putting them together.

I like and want to write more, and then I really enjoy coding and being entrepreneurial. So I think being able to put all that together is for me the sweet spot. So I enjoy writing and I also enjoy creating writing tools quite a bit.

Steph Pajonas: I’m with you on that one. I am also slowly working through a book right now.

It’s like I’ve finally, I think, hit the midpoint of it, but I’ve been working on it for months because I also really enjoy helping other authors, which is how, which is why I do this podcast, which is why I’m part of FFA, why I write my Substack, these sorts of things, because I’m, I want to help authors. I see AI as a huge market disruptor.

It’s a huge market disruptor, and it’s going to disrupt not just creative industries, but it’s also disrupting everything else as well. [00:16:00] Between, like, the amount of data centers we’ve got to put in, the corporations who are using it on the back end in order to process data, the medical breakthroughs that we’re having right now.

It is just all over the place. It is not just affecting creatives. But since I’m an author and I’m a creative, I wanna help those people get through this huge disruption in our industry. So I’m also, like, like Joe, and I like to code as well, so it’s like I, I totally understand Joe and where he’s coming from.

We are not always able to have a ton of extra time to write on the book, but it’s nice to have it there, and you can go back to it if you need to, and that’s what I do. When I’m like, “I need a break. I’m gonna go work on my book,” right? So yeah, it, I can see Joe definitely keep, keeping on working on his book and then working on more as time goes on because we have a lot of tools that we’ve been making that have been helping with a lot of this stuff.

Joseph Ucuzoglu: Yeah, and the market [00:17:00] tools are getting amazing. Like with … I feel like right now with Claude Code, it … You always had the ability to do the, vibe coding. You’ve heard that term, running around for a couple years. But I think it’s recently that the vibe coding has gotten to the point where just somebody with an idea can sit down and just tell the computer, “This is what I want,” and then it’s actually gonna execute it in a way that is, is gonna work.

Before, people could stitch together little things but now I feel like we- we’ve gotten over the hump. I feel if you’re a person you don’t even have to be a coder. If you just wanna be able to create your own apps for your own purposes, right? Let’s say you wanna have a little app to, I don’t know, simulate note cards on a corkboard for your writing.

That is within the capabilities of anybody who learns the tools and just describes what they want. They don’t have to know how to code anymore.

Danica Favorite: I really love that because as I’ve said multiple times on the podcast, I am not the coder. [00:18:00] I am not a computer nerd. You should see… it was funny, you were talking about the old Rexy and everything, and I still have all of that in my Notion. My Notion is still a hot flaming mess, and, you know, today I went in there and I said, “Okay, please update,” ’cause I love it. I’ve got the Notion with AI, and I said, “Okay, I’ve just put all these articles on my trip in here. Please organize this for me.” And it did it. So I’m very happy about the fact that I don’t have to know any coding or figure that out. So what are some tips that you might have for someone like me and some of our listeners who are not the coders? Like, where would you suggest we start in playing around and figuring out some coding stuff that could help us with our writing?

Joseph Ucuzoglu: Sure, yeah. So this is something I’m- I’m hoping to do a full course on for the FFA in the future. But for right now, what I would advise people to do is you need a pipeline to be able to get yourself from the point where you have your [00:19:00] idea to the point where you can deliver on it, okay?

My favorite pipeline involves either the desktop version of Claude Code or codex from ChatGPT. Both of those allow you to just describe what type of app you want. They work with local files on your hard drive, and they’ll create a working program that you can test locally.

Then I also like to work with those tools along with another IDE. That IDE could be Cursor, it could be Antigravity, it could be Visual Studio Code. For me it’s Cursor. And that tool makes it easy for me to review the code that’s been developed by Claude Code or Codex, and it makes it really easy for me to use a connection to GitHub.

So they have this little module that’s built into all of those tools that makes it really easy to just see what files have been edited, and then you just click one button, and then it’ll push it to this [00:20:00] repository. And so this repository’s just a cloud storage place. Stores your code.

It’ll keep track of the versions. That’s not really the most important part. I just like it as a bridge because then what I also have is there’s a service called Railway, and that service allows you to just spin up these applications, these little containers, and they’ll look at GitHub and see, like, oh, you just updated this.

Let me go ahead and redeploy your code. And it also provides a URL for you, so now you have something that you can use on your machine but also share with everybody, share with the rest of the world. So literally, once you learn that pipeline of kind of those tools and just the simple methods to push the data over, you have everything you need.

And used to be that there was some complication to that putting stuff on Railway, you would have to learn the nuance of certain configuration files and other things. And now that doesn’t [00:21:00] really exist. You just tell Claude Code, “Here’s what I’m planning to do. Can you just create any files that are necessary?”

And it one-shots it, and it creates those files, so then it just works when you push everything. So there’s only a small amount of information, I think, like gaps that need to be bridged before just anybody can copy that pipeline. And then once you have the pipeline down, then you’re off to the races, and you create whatever you want.

Steph Pajonas: I was just telling my husband this morning that, um, it used to be that there was like a huge moat between you know, normal, regular people and programmers who knew all of this language, right?

They could get on a command line, and they’re typing away, and they’re using all these commands, and they’re doing all this cool stuff, and I was just thinking about, like, all the Hollywood movies that would get the hacker in there that would do all the crazy stuff, and he’s like hacking in all these things.

And nowadays you can just go tell Claude to go do something, and you [00:22:00] can do it in plain language, right? You don’t have to know a command line code. You don’t have to know how to do PHP. You don’t know how … You don’t even know how to do Python or anything like this. You can just go ask for it in plain language, be like, “I want a dashboard that’s going to show me my word counts every day,” and Claude will just build it.

Claude will just build it for you and show it to you right there in chat. And then there are ways also, like Joseph was saying, about getting it onto a server so that you can access that from anywhere. That’s a great idea, is to have a little pipeline going from, Cursor, which is an IDE.

We’ve talked about these IDEs in the past on the podcast, and going to GitHub, and GitHub becomes a place where Railway can pick it up and put it on their own servers and then serve it up to the public. So I mean, I, I agree with this because I think more and more people are going [00:23:00] to want some sort of web presence that doesn’t necessarily have to be WordPress.

In the past, if you were an author, before AI, I would have said go get hosting, get a WordPress instance going, figure out what kind of templates you want, and then put together a website.” And then for some people that was just beyond them because of the fact that WordPress is, WordPress can be sometimes hard to use because it has its own vocabulary. There are plugins, there are themes, there’s all … And then you have to harden it against being attacked because it is one of the biggest content management systems out there, so everybody loves to hack it. So anyway, I was just helping a friend with that the other day.

So- in the past we had to do that, but now you could go ask Claude for a static website that doesn’t have a database. It’s not gonna be hacked. It’s pretty hardened, and you can just have it deployed to a website. And then boom, you’re set to go. So I love this fact that we have these tools now.[00:24:00]

Joe’s been working with them for quite some time, and he’s been building lots of tools for us as well, so that’s great too. We don’t necessarily have a lot of time to build, put together both curriculum and to put together tools, so Joe’s our tool guy. He’s really good at that. But he’s also teaching some curriculum at the FFA, so I would love for you to talk about some of the stuff you’re putting together there as well.

Joseph Ucuzoglu: Yeah, so the idea is gonna be for to, with the Ink & Code to take authors and empower them to be builders in terms of whether that’s with your website, whether that’s with little applications that do things. ‘Cause again, it’s kinda if you as an author could imagine being able to just have just whatever amount of money you need to hire your own coder to personally follow your instructions to build what you ask for, that is pretty much where we are.

That … you need to know a little bit about [00:25:00] how to kinda connect the pieces together in a way that will allow you to take what the LLM gives you so that you can make it available at large. I’m sure they’re gonna make that process even easier probably in another six months. But for now there’s just little bits that you need to know that kind of bridge that, and then you can create your own tools and so forth.

So that’s, where I’m focusing is just looking at the overall tools that are out there and how we can use them and how authors can, you know, take that information and, do more with it. And so that’s also too with things like Claude Cowork, creating skills and things like that.

There’s just so many ways with the tools that are out there to do things. And it’s not even just with words, right? You look at ChatGPT just came out with their Image 2 yesterday. That thing’s crazy. You can literally go in there if you have a paid account and this also works with Nano Banana, from Gemini, although the [00:26:00] ChatGPT one is better.

And you can take your chapter and just say, “Hey, I want you to create multiple pages for a graphic novel from my writing.” And, I’m amazed when we did this, like, you would see it. It would create it. It looks really good. And then things are consistent. This is what amazed me.

I remember doing this with a story, and then I saw the person leaving their house, and it had a couple of bushes or whatever, and they go off and they do something, and then they come back later at night. And when they came back to their house, it still had the two bushes in the right place, and that amazed me.

‘Cause if you know from the earlier days with image generation and stuff, you probably leave that house and you come back and you’re living in a mansion or something. Things would just not stay very consistent.

Danica Favorite: And I think even like when you look at like characters and things like that, like that’s been the toughest thing.

I actually I did just bite the bullet and hired a cover designer on the series that I’m working on, because originally I had an AI cover, didn’t [00:27:00] love it, sent it to someone, they fixed it and made it pretty for me. And then I was like, “Okay, but now this is gonna be a series, and I don’t know that this cover is the right vibe for a whole series.”

Which until yesterday you couldn’t really get with a consistent thing unless you did a lot of playing around and everything, and now that’s fixed. I’m still having the cover designer do it because as we’ve said in the podcast as early as our last episode, you know, where it’s like, it’s either gonna cost you time or money.

Everything’s got that cost. And at this point I’m like, you know what? For this project, I’m spending money rather than spending time. And I like even what you’re talking about with the different ways that people can build their systems and build these apps and things like that, is if you want to take the time to learn it, it’s possible even for someone like me who doesn’t have a lot of skill.

Or it now becomes [00:28:00] cheaper for me to go to someone and say, “Hey, can I pay you to build this for me?” Because that’s gonna take them less time and less energy. And so it’s a win-win on all sides because maybe I will never develop the skills to do this effectively. But somebody in the FFA or elsewhere is going to have the skill to be able to do that for me and do it easily without spending hours and days and months and all of that.

So I think it’s super exciting that these possibilities are all coming.

Joseph Ucuzoglu: Yeah and for draft sort of stuff, it’s really good. You can take your cover that you have right now and drag it into something like ChatGPT with this new Image 2, and say, “Hey, give me, five variations,” and let’s say those five variations are gonna be, my sequels, and here’s the names and the titles of those, and a little bit about what they’re about.

And you can get some things that look really consistent and good, and maybe they won’t be what you can just run to market with, right? You [00:29:00] may still have to go back to, a designer or something like that. But at least you can be a lot closer when you go back to them and say, “Hey, look, I’ve got these. I kinda like ’em. Here’s where they’re kinda not doing what I want them to do.” And I th- I think it’s a huge benefit, and it’s just fun. It’s just so fun to be able-

Danica Favorite: Oh, yeah.

Joseph Ucuzoglu: … to see things.

Danica Favorite: Absolutely.

Joseph Ucuzoglu: You can visualize your characters now, and that just takes a request, and then now you’ve got pictures to inspire you or to put up on your mood board or whatever you have.

Steph Pajonas: It takes some ambiguity-

Joseph Ucuzoglu: Yeah

Steph Pajonas: … out of the process, right? So you’re able to go in there with a little bit more, with some more ideas about what you want, what you’re interested in, and then you’re able to communicate better with the other person, which is great. I love having that now as a tool in my back pocket when I need to go and explain myself.

I can use visual aids or whatever it may be, which is great. I wanted to touch on the whole software, because we produced a lot of software at the FFA for, like, the [00:30:00] last two, you know, plus years. We’ve had Rexy. We’ve had Your First Draft. We’ve made a lot of internal software, and one of the cool things that we realized in the last six months was the fact that if we can teach authors how to build their own software, they don’t need to be dependent on a third party to give them what they want.

So I was just going to focus on the fact that we’re teaching people how to do these things in order to empower them and learn how to do the parts that, make make applications as part of their process for their business, right? So one of the things that Joe’s gonna be teaching people is how to get started with that and that is really exciting for me, because I wanna see more people be empowered to do those kinds of things.

Joseph Ucuzoglu: Yeah, with these tools it really does make it possible for someone to essentially get [00:31:00] the same sort of output that they’re getting from our tools right now but with something that they create. That wasn’t possible a couple years ago when, you know, vibe coding started. You still had to have a much higher level of technical proficiency to be able to, you know, get something useful out of it.

But nowadays, I feel like if you have somebody who can tell you “Here’s some things you can do,” and show you some patterns, just with writing, like some people can show you some prompts too that can help with certain things. You can produce your own code.

And the bar to produce something that is personally useful i- is a lot it’s a lot lower than what you would have to do if you need it to be used by a lot of different people, and if it has to be kind of something that’s more commercially viable. Because when you’re doing it for yourself, you can update things.

You don’t have to worry about the server being overloaded or anything like that. So i- it’s really nice because some of these things oh, you have the sequencer through YFD, and, and it [00:32:00] does a lot, right? But if you yourself wanted to create something to create your own I don’t know, detective noir novels that you’re creating, like, miniseries, to create a tool that just does that is, is definitely within the reach of any author, I would say, who just wants to sit down and just learn some of these tools to be able to create it, because the tools are that powerful now that they will help you create your own software

Danica Favorite: I think that’s so exciting because, like everyone has their own process and we- we’ve talked about this a little bit in the past where I personally love seeing other people’s prompts and learning from their prompting and taking them. But for me, there’s always something in their prompt that I tweak.

A- and that’s exactly what you should be doing. You should, you know, like people are like, “Oh, just give me the magic prompts and I’ll make this thing.” And then they’re disappointed ’cause they don’t get the results. And I think that’s also gonna be true with the software and the different software [00:33:00] systems because, how many of you have tried something like Sudowrite or Novelcrafter or some of the other softwares out there and been disappointed in the results because you don’t get to see the prompt, you don’t get to control the prompt and change things around, which is… I mean, that’s why the FFA got started. Because Elizabeth believes very strongly in creatives controlling the prompts and seeing the prompts and seeing the models. And I think for a lot of people, that was not accessible for a long time because they didn’t understand it. And now it just blows my mind how different that’s becoming that really anybody who wants to use these tools who’ve been saying all along, “Oh, I can’t afford this, I can’t afford that,” guess what?

Like now the ability for you to learn how to do it and do it on your own and do it well it’s here.

Joseph Ucuzoglu: Yeah, definitely. I- it’s amazing how far we’ve come, right? We started with the, you know, creating sentences, then [00:34:00] stringing paragraphs together. Then, we had tools like RaptorWrite, that went through and allowed you to control the context and turn things off and on easier, so it was much faster than copying and pasting.

And that allowed us to produce, like chapter-sized outputs I would say using different methods. And then from there we went to full sequencing with things like, YFD. And now when you’re looking at things like, with Claude Projects and Claude Cowork and a lot of people doing things in Antigravity now there are ways…

Like, people have already put together their own kind of different types of variations on, essentially what YFD is doing. And a lot of them have managed this on their own. And so that’s where we came up with the idea of oh we should be, helping even more authors be able to do this if they want.

And since then the tools have gotten more powerful, so the, the processes have gotten even easier. So it’s just like now it’s gonna be about showing people “Hey, this is what you can do, and here are some of the [00:35:00] tools that if you know how to use these specific tools will allow you to accomplish these sort of tasks.”

But yeah I’m amazed. We have this thing within the FFA, we call it the Show Joe Show, where … ‘Cause, people like, used to like to show me some of the stuff that they would make, so we just made a whole show out of it. And, I am always amazed at what comes out from these authors.

Extremely impressed. Like, as a, you know, as a person who’s a coder and I’m watching authors going through … And then some have a little bit of a tech background too, but just what they’re producing is just you know, jaw-dropping, sort of stuff that they’re putting together.

So it-

Danica Favorite: That’s great. I love that. That’s amazing. So what are you working on now?

Joseph Ucuzoglu: My big push right now, I’m working a lot on the new pro membership that we’re gonna be coming out with over at the FFA. And then from there, I am switching gears into digging into Ink and Code and just really empowering authors to, to build their own [00:36:00] tools.

So on the side, if you’re asking me as a side project, like what might I be doing in the same space outside of FFA, I’ve been coding an application that I’m using. It was that notecard app I had mentioned earlier. So this idea of I have these notecards, I wanna be able to put text on them and images and move them around on a canvas, but then I want to have the large language model there that can construct scenes on these cards because when I like to think through a story I just have these ideas for the tent pole events of okay, this is a really cool idea.

This is an idea. These are the things I like. And so I wanna throw them all there, but I find it really difficult to do in just like a flat out just writing software like Raptor Write or something like that, just to have these kind of random, fragments. So to me, if I have the random fragments, I want something equivalent to having notecards that I can move around.

So I made the actual visual of that. And it’s awesome because I am terrible at front end [00:37:00] UI, kind of JavaScript, being able to move things around. Even though I have a programming background, I’m terrible at that, but it doesn’t matter today, right? I was able to just tell the, tell Claude Code, “Hey, this is what I wanna do. Here are some of my ideas. You know, I would like it all to maintain inside the client of the browser so that I don’t have to spin up a server or anything like that. What do you think of my idea?” So that’s usually what I’ll do is I’ll vet my idea and I’ll ask it. And it’ll come back and then, it always tells me, “Hey, this is an amazing idea,” even if it’s not amazing.

But it will then put in some information about, these are some opportunities that you have to maybe fix it, or maybe this would work a little bit better. And then at that point, it’ll ask me, “Hey, do you want me to whip this up?” And I say, you know, “Yes, go ahead and do it.” And then it puts something there, and then I can see it, and I can see okay, this part I like, this part I don’t.

And then I can just iterate just through standard language. I’m not even looking at the code. I used to look at the code when I did Vibe coding. I used to look at every line of code that came out to make sure it was okay. Now I’m finding that even as a coder, I don’t look [00:38:00] at the code. I just don’t look at the code.

I just ask for it, I get it, and then I push it, but I don’t look at it. Not very often.

Steph Pajonas: Me neither.

Yeah.

Especially now that it can write to five, six, 10 documents, like, all at once. It’s like I don’t even necessarily know where it’s putting new code every time I ask for something, right? But I will have to open up a specific extra document that shows me its implementation plan and the steps that it took.

I can actually go back and look up through the conversation and open up the thinking section so that I can see that it was like, “Oh, I think I need to put this in styles, and then I need to add this thing to the JavaScript,” so I can see where it went and did things. But I don’t proof it as much anymore because Claude has just gotten that good and so has GPT Codex now.

They’re at the point where, where, where Anthropic is hiding models from us because they’re so [00:39:00] good at code that they could probably break into anything, right? So we’re now at this point, this tipping point when it comes to code, where the models can do just about anything, and you can ask for it in natural language.

You don’t have to say, “You need to build this Python script to do this, that, or any other thing.” You can say, “I want this thing,” and it will go off and it will make that decision about which programming language, those sorts of things, how it’s going to structure the files, all of that. You don’t have to do that anymore.

I love that. It’s so great because it just takes a huge load off your shoulders and allows you to iterate like what Joe’s doing because he will go… he’ll work on his project, and he’ll come up with a new idea and then present it to the AI, and it will go off and execute that idea, right?

Joseph Ucuzoglu: Yeah it’s getting really neat.

I would say about 30% of the time, like I’ll go through and I’ll ask it, how it wants to do it, ’cause I like to have some [00:40:00] control over the design approach for what it’s gonna take. And I would say about six or seven out of 10 times I will have my specific approach that I’ll want it to take, which may or may not be, what it’s chosen.

Not that it chooses bad, but I just wanna do something that I think is a little bit better. Maybe it wants to roll something up and I’m like, “No, I think you should just use this library I know about. I think that would be a little bit more efficient.” But what I’ve found is in a lot of cases, and it’s happening more and more often, and it’s scary, is that I would say three out of, two out of three, two or three out of 10 times it’ll come up with ideas, and those ideas, some of them will be better than I know I could have thought about or come up with myself if I had just sat there i- in the afternoon just, thinking through, how am I gonna do this?

And it will choose the best one of those ideas. It’s … The way it’s ranking the ideas it knows from my perspective which one is probably the best [00:41:00] decision, and it’s almost always choosing that. So I would say 20 or 30% of the time it’s making superior design decisions to what I would make as, somebody who’s been doing this for, 15 years, 20 years.

And so it, it makes me think that it’s not gonna be too long until I can probably trust it to make, in a lot of cases, better decisions about developing these apps than I can. Which means that for as an author, we’re not that far from being able to just have a little bot that we can say, “Hey, I need this program,” and then, do this research because this is what I’m looking to do, and create this program for me to help me write or whatever process you want it to do.

And it’s gonna be able to go out there and just do that process for you without you having to help like you have to now with the pipeline, right? You won’t even have to help. It’ll just do it, come back to you, you give it feedback probably through your phone, right? You probably won’t even have to be sitting there [00:42:00] anymore.

And then, eventually you’ll just have these tools, things that, it would take people months to build, built for you, while you’re drinking your coffee.

Danica Favorite: I am so looking forward to that. Um, and you know, it was, it was great. Even this morning, I was in a meeting and I had a whole Claude thing just running in the background, and I’m just sitting here doing my thing. I don’t have anything going now ’cause it’s recording and I don’t want it to break my poor little baby computer with all that.

But it is nice because you can just set it, “Here, do this thing,” and you’re off doing something else, which is so nice. So as we’re wrapping up, I wanna talk about favorite AI tools and what your favorite is, but I wanna give you a little bit of a twist because we’ve talked about a lot of tools.

So do you have a place of tools that, number one, let’s talk about your personal favorite, but I also would love to hear some recommendations for maybe someone who isn’t that experienced with coding, where you would recommend they start or a good first tool or [00:43:00] something like that. But please also tell us about your favorites

Joseph Ucuzoglu: Yeah, I think right now the ones I’m really liking, if we’re talking about coding a- and even, working on some story elements, I really enjoy the Claude Desktop and kind of ChatGPT’s desktop, the Codex.

Those two I think are really nice. They work really well, and they’re really powerful and you can do a whole lot with them. That’s what I would say. If somebody was going to be starting as an author, I would just say Claude Desktop, right? That one, it just has so many features in there.

The only thing it’s missing really is the image stuff, right? They didn’t, they really stayed out of the image game. And for that, I think, which I think it’s great. The way that the new images from NanoBanana and from chatGPT’s image model, what they can do in terms of taking content and creating these amazing infographics or, taking your stuff and making it into, you know, some graphic novelizations and other things, it’s just super fun.

[00:44:00] So if it’s from an image perspective, it’s gonna be ChatGPT and Gemini for the NanoBanana. And if we’re talking about just general productivity and coding then we’re talking Claude Desktop and Codex on my machine, ’cause there is a cloud version of it, but, having it on your machine makes it much more powerful.

And it’s getting more powerful every day. It’s almost to the point now where i- in, and these, some of these features are available where they’ll actually take control of your computer and do some tasks for you.

Steph Pajonas: That’s what I want Claude to be doing. Come do my work for me, Claude. Click on these links, download these things.

That’s what I need Claude to be doing right now. That’s awesome. I’m glad that we are able to give people a place to start, especially with some of these tools. There’s also plenty of stuff on YouTube e- not even just Future Fiction Academy, although we do have plenty of videos on YouTube talking about coding, talking about the way we use these tools.

[00:45:00] We have some Ink and Code videos on the FFA channel too that might interest people if they listen to this podcast and they’re like, “Oh, I kinda wanna go check that out.” They could go check that out on the Future of Fiction Academy YouTube channel. But also I want to just point out that there are plenty of YouTube creators as well that are just showing the basics about these things, like what is GitHub?

You don’t necessarily have to go and have somebody- teach that to you. You can go watch a few YouTube videos on it, get familiarized with these tools if you decide to do Cursor or Antigravity. There are plenty of tutorials on those as well on YouTube. So like Joe has given us a great start there, so I’m hoping that people will want to go check that out when they listen to this.

And so before we go, I wanna make sure that we give anybody any kind of URLs or where to find you, Joe, besides the Future Fiction Academy. [00:46:00] Where can people find you? Oh,

Joseph Ucuzoglu: The best place is to find me is the Future Fiction Academy. Yeah, so there’s no special page for me where I’m advertising myself or anything like that but yeah-

Steph Pajonas: Perfect

Joseph Ucuzoglu: just come on down, say hello if you’re in the FFA. I’m always there in the Discord. And then I just wanted to throw out there too, you were mentioning some of the YouTube stuff. I think i- if people are interested in the slightly bigger tech dives, Matt Wolfe and Matthew Berman. Those are both interesting ’cause they always talk about what’s the latest happening in, in AI, and I usually think it’s pretty easy to see that and then see oh, here’s how this could connect to doing something and writing. Or something useful for authors.

Steph Pajonas: Yeah, I love Matt Wolfe too. I watch his videos every week to get caught up on what’s going on in AI. So that is a definite must-watch every single week. Thanks for that. I totally forgot about that. So I’m going to put those links as well as any other links that we talked about tools we talked about, in the notes.

[00:47:00] Anybody who’s been listening, come by bravenewbookshelf.com, check out the show notes for this particular episode. Like I said, I’ll put everything in there for us, as I usually do. And you can subscribe to the show notes by hitting subscribe at the top of the page. Come by, click it, give us your email address.

We’ll send you the show notes the day after the show goes live. Danica, anything else we wanna talk about?

Danica Favorite: I really did appreciate the reminder of where you can find the free information, because I think that we all feel very passionately about the fact that we want people to be able to do and learn AI at whatever stage financially you are.

And so please check out those free resources. And I know FFA’s got some good free resources. Publish Drive also has some ways to start free. And we have been very remiss, and we’re updating all of this, but please take a look at the sponsor page on the website because we do have special [00:48:00] deals and offers from both FFA and Publish Drive for our listeners.

So make sure you take advantage of that. And also remember to like and subscribe to Brave New Bookshelf on Facebook, YouTube, and then all of the socials for FFA, Future Fiction Press, and Publish Drive Um, and yeah, follow us everywhere because we are all giving a lot of really great information. And if there is a topic that we haven’t covered or something you’d like to see from any of us, particularly as we’re winding down the season and I’ll start looking to book new guests, make sure you reach out.

Send us an email and let us know what other things you want us to cover, talk about, share, because we do want this to be a very good resource for everyone.

Steph Pajonas: We wanna hear from you. So come by and either leave a comment or you can hit us on one of the social media channels. You could also leave a comment on YouTube on one of the videos if anything struck your [00:49:00] fancy and you thought, “Hey, I really wanna hear more about that,” let us know.

We’re excited to talk to more people about all of these things, definitely. So thank you again, Joe, for being here and talking with us today. Really appreciate it. Everybody else, thank you for being here, and we’ll see you guys in the next week’s episode. Okay,

Bye.

Danica Favorite: Bye.

Joseph Ucuzoglu: Bye.

Speaker 2: Thanks for joining us on The Brave New Bookshelf. Be sure to like and subscribe to us on YouTube and your favorite podcast app. You can also visit us bravenewbookshelf.com, sign up for our newsletter, and get all the show notes.

Episode 82 – From Writer to Developer: Vibe Coding and Building Your Own AI Tools with Joseph Ucuzoglu