← Back to search

#704 Neil: AI Side Hustle Ideas You Can Sell With Claude Fable 5.2

AI Fire Daily · 2026-11-16 · 14 min
relevance 53 2495 words Episode page ↗ Audio ↗
Show full episode description
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Learn how to build and sell AI Side Hustle projects with Claude Fable 5.2. See how to create premium 3D websites, playable game ads, and custom talking storybooks, plus explore 3D property tours and an AI agent workflow for finding freelance jobs and preparing client proposals. 💸 We'll Talk About: Building premium 3D websites with Claude Fable 5.2. Creating interactive HTML5 playable game ads. Selling personalized talking storybooks. Creating 3D property walkthroughs with Higgsfield and Kling 3.0. Using Hermes Agent and Claude Fable 5.2 to find freelance jobs and prepare proposals. Keywords: AI Side Hustle, Claude Fable 5.2, AI Side Hustle Ideas, Playable Game Ads, HTML5 Playable Ads, AI Tools. Links: Newsletter: Sign up for our FREE daily newsletter. Our Community: Get 3-level AI tutorials across industries. Join AI Fire Academy: 500+ advanced AI workflows ($14,500+ Value) Our Socials: Facebook Group: Join 303K+ AI builders X (Twitter): Follow us for daily AI drops YouTube: Watch AI walkthroughs & tutorials
0:00 / 0:00
[SPEAKER_03] A 3D interactive website can easily be listed for $250 online. But building the demo with an AI tool like Claude Fable 5.2 costs around $5. Think about that margin for a second. [SPEAKER_00] Yeah, it's a wild economic arbitrage. It kind of completely changes the freelance game. [SPEAKER_03] Welcome to the deep dive. We've got a very specific mission today. We are moving past that generic advice of just making money with AI. You see those videos everywhere, right? [SPEAKER_00] Oh, definitely. The endless tutorials about faceless YouTube channels or automated blogging. [SPEAKER_03] Exactly. We want to deconstruct five practical AI side hustles for you. We are detailing the exact costs. We are looking at the actual build times. And we are breaking down what you can actually charge clients. [SPEAKER_02] Because the execution is what actually matters here. Ideas are incredibly cheap right now. Building the tangible thing is where the real value lives. [SPEAKER_03] Okay. Let's unpack this. Look, I still wrestle with prompt drift myself. It can be super overwhelming when the AI just slowly forgets your original instructions. But the workflow we are exploring Today is genuinely different. [SPEAKER_01] Mm-hmm. You aren't just staring at a blinking cursor. You aren't writing everything from scratch. [SPEAKER_03] Right. You are using AI alongside existing templates. It is kind of like stacking Lego blocks of data. [SPEAKER_02] That's a perfect way to look at it. You are essentially an assembler. You aren't acting as a raw [SPEAKER_03] manufacturer. We need to move from the theory of AI arbitrage to the first practical application. Let's look closely at high-end web design. [SPEAKER_02] Right. Brands and beauty, fashion, and luxury have a very specific problem right now. They desperately need premium shopping experiences to stand out in a crowded market. [SPEAKER_03] They need their digital storefront to match their physical luxury branding. [SPEAKER_02] Exactly. Imagine a hyper-realistic 3D perfume bottle sitting on a screen. It rotates slowly. It physically follows you as you scroll down the page. [SPEAKER_03] Yeah, that creates an immediate visceral reaction. It instantly feels expensive and bespoke. [SPEAKER_02] And building that used to require a whole team of specialized software developers. It took weeks of custom coding. Now you can offer this exact service on Fiverr or Upwork as a solo freelancer. [SPEAKER_03] Here's where it gets really interesting though. You don't just ask Claude to build a website from nothing. You actually go find an existing open source project on GitHub first. You load that directly into Claude Fable 5.2. [SPEAKER_02] You let the open source community do the heavy lifting for you. [SPEAKER_03] You're leveraging the existing code base as your structural foundation. And just to clarify for everyone, a code base is just the complete set of instructions making software run. [SPEAKER_02] Right. You use that foundation and you prompt the AI to customize it. But the prompt instructions need to be highly specific. [SPEAKER_03] You can't just say, uh, make it look cool. [SPEAKER_02] No, not at all. You ask for smooth frame rate animations. You demand scroll-based visual triggers. You include a functional add to bag button. And it absolutely has to be mobile responsive. [SPEAKER_03] But wait, if I'm not a software engineer, how am I actually doing this? Am I just hoping the AI guesses my aesthetic taste? [SPEAKER_02] You're acting as an art director. You tell the AI to make the bottle look realistic and keep the lighting natural. You test the interactive sections in Claude's preview window. [SPEAKER_03] So you spot the visual bugs yourself. [SPEAKER_02] Yeah. You find a glitch and you just tell the AI to fix it in plain English. You refine the code until the shopping experience is completely polished. [SPEAKER_03] So let's talk about the raw economics of this hustle. The margins are pretty compelling for a beginner. [SPEAKER_02] They really are. You can sell a custom hero section on Upwork for around $90. A hero section is just that big visual header at the top of a site. [SPEAKER_03] And if you're building full sites? [SPEAKER_02] If you build full interactive sites on Fiverr, that scales from $250 up to over $1,000 per project. [SPEAKER_03] That's a serious return on a $5 compute cost. [SPEAKER_02] It's massive. A great strategy for beginners is offering structured pricing tiers. You create a $250 starter package. That includes a one-page site with scroll animations delivered in four days. [SPEAKER_03] And then you offer an upsell for the impatient clients. [SPEAKER_02] Exactly. You offer a $400 express package. It's the exact same customized website, but you deliver it in one day. [SPEAKER_03] That brings up an obvious hurdle, though. How do you actually stand out on these crowded freelance platforms to get that very first client? [SPEAKER_02] You do the work before you get the job. You build a small, customized working demo of their specific request. [SPEAKER_03] You build it using their brand colors and their logo. [SPEAKER_02] Yes. [SPEAKER_03] Yeah. [SPEAKER_02] You attach that customized link to your proposal. [SPEAKER_03] Before you even hit submit. [SPEAKER_02] Yeah. [SPEAKER_03] So, showing a working prototype up front completely bypasses the standard resume filter. [SPEAKER_02] Right. And it proves you can actually execute. You aren't just making empty promises to a nervous client. [SPEAKER_03] So, we have solved visual attention with 3D websites. But luxury brands just want you to look. Mobile game studios have a much harder problem. They need you to physically interact with their product. [SPEAKER_02] It takes digital engagement to the absolute next level, which leads us perfectly into the mobile gaming space. Mobile game studios have a massive user acquisition problem right now. [SPEAKER_03] They need ads that let users actually play a tiny snippet of the game. [SPEAKER_02] Before they even commit to downloading it from the app store. Nobody really wants to watch a static video ad for a mobile game anymore. [SPEAKER_03] Right. So, instead of building a massive full game, you use AI to build a playable browser ad. [SPEAKER_02] Let's look at the Bubble Harbor example from the source material. It's a fictional mobile game concept used as a test case. [SPEAKER_03] So, walk us through the prompt workflow for this game ad. [SPEAKER_02] You ask Claude to build a colorful bubble shooter. The core gameplay loop is intentionally simple. You just drag a finger to aim, match three colors, and clear the board. [SPEAKER_03] And then you add the psychological hook at the end. [SPEAKER_02] Yes. The win state automatically triggers an install now button. You specify a coastal harbor theme in the prompt. You ask for smooth bubble animations and highly satisfying popping sound effects. Hmm. [SPEAKER_03] Wow. Imagine scaling to a billion queries. Or, in this case, a billion ad interactions from a single prompt. [SPEAKER_02] It's totally staggering when you really think about the sheer scale. Millions of people playing your code. [SPEAKER_03] And the technical requirements here are very strict. [SPEAKER_02] Very strict. You prompt the AI for HTML, CSS, and JavaScript only. Absolutely no static images faking the actual gameplay. [SPEAKER_03] It has to run directly in a mobile web browser. It has to be incredibly lightweight to load instantly. [SPEAKER_02] Let's share the hard numbers on this specific build. This playable ad took just 26 minutes to generate from start to finish. [SPEAKER_03] 26 minutes to build a functioning, interactive game slice. [SPEAKER_02] Yeah. And it cost about $6 in API compute time. [SPEAKER_03] Just to clarify for everyone, API compute time just means the tiny fraction of a cent you pay OpenAI or Anthropic each time their brain processes your request behind the scenes. [SPEAKER_02] That's spot on. It's purely the hidden processing cost. And the market value for this asset is fantastic. You can charge mobile studios between $300 and $1,000 per project. [SPEAKER_03] I'm curious about the underlying mechanics here. Why wouldn't a mobile game studio just use a pre-recorded video ad instead of paying a freelancer for this? [SPEAKER_02] Because playable ads drastically increase the actual conversion rates, the user has already invested physical effort into tapping the screen. They've actually experienced the gameplay firsthand without ever leaving their current app. [SPEAKER_03] Right. Letting them test drive the dopamine lube guarantees a much higher download intent. [SPEAKER_02] Exactly. [SPEAKER_03] You're selling the studio higher conversion rates, not just raw JavaScript code. Mid-roll sponsor read placeholder. [SPEAKER_02] All right. Let's keep going. [SPEAKER_03] Okay. We have looked exclusively at B2B services up to this point. We are selling to luxury brands and mobile game studios. But AI's ability to generate custom code is equally lucrative when we pivot to B2C. [SPEAKER_02] Selling directly to everyday consumers. [SPEAKER_03] And more specifically, selling directly to eager parents. What's fascinating here is the application of deep personalization. We are talking about custom talking storybooks. [SPEAKER_02] Let's break down the actual workflow for this one. You target parents of kids aged roughly four to seven. You are using the Clawed Code app here. You prompt the AI to build a customized web form. The web form collects very specific narrative inputs from the buyer. The parent enters the child's first name. They enter their current age. [SPEAKER_03] They pick their favorite animal and their absolute favorite color. [SPEAKER_02] And they pick an overarching story theme. Something like space exploration or magic forests or maybe dinosaur friendship. [SPEAKER_03] The AI takes all those unique data inputs. It dynamically generates an eight-page illustrated storybook. [SPEAKER_02] But it's not just static text on a boring screen, though. It has smooth page-turning animations built in. It uses browser text-to-speech for the actual audio narration. [SPEAKER_03] It even highlights the current page's text while the narration plays along. And there's a seamless PDF export option, so they can easily print it out. [SPEAKER_02] So how do the economics of this actually work? You clearly aren't pitching a bedtime story to corporate clients on Upwork. [SPEAKER_03] No. You've set up a storefront and sell these directly on Etsy. You can list them as custom digital products. [SPEAKER_02] The absolute secret sauce here is the underlying scalability of the code. You essentially reuse the exact same application for every single customer order. [SPEAKER_03] Precisely. You merely change the form inputs based on what the parent typed in at checkout. The app instantly generates a brand new, entirely unique digital product. [SPEAKER_02] Yeah. You just spin up a new version in seconds. [SPEAKER_03] I have to ask, though. Aren't there already thousands of cheap children's books out there? Why does this specific digital format sell so well? [SPEAKER_02] Because the child actually hears their own name read aloud by the browser. It physically makes them the actual protagonist of the ongoing story. [SPEAKER_03] It transforms a passive reading experience into a highly personal, interactive digital adventure. [SPEAKER_02] It's pure emotional value. Parents will always gladly pay a premium for that kind of personalized magic. [SPEAKER_03] Up until now, we have used AI to build the actual deliverable product. The 3D website. The playable game ad. The personalized children's book. [SPEAKER_02] Right. We focused entirely on generating the deliverable itself. [SPEAKER_03] But what if you could use AI to completely automate the client acquisition process itself? [SPEAKER_02] This is where we get into the bonus hustles. If we connect this to the bigger picture, we are automating the tedious parts of the hustle itself. [SPEAKER_03] Let's look at the first bonus idea. 3D property walkthroughs for the massive real estate market. [SPEAKER_02] This workflow combines a few different powerful AI tools. You use Higgs Field and Kling 3.0. You just take standard flat property photos. [SPEAKER_03] Just the regular, boring, real estate listing photos you see online everywhere. [SPEAKER_02] Yeah. You feed those static photos into the AI. It calculates the depth and turns them into realistic, cinematic 3D video tours. [SPEAKER_03] It basically fakes a drone flying through the living room. Then you use Claude to write the final presentation pitch for the agent. [SPEAKER_02] The profit margins on this are genuinely great. It costs under $10 an API compute to run the video generation. It takes about 40 minutes of your time to oversee. [SPEAKER_03] And you sell that final polished package to real estate agents for anywhere from $150 to $450. [SPEAKER_02] Because every real estate agent desperately wants their property listing to look cinematic and premium on social media. [SPEAKER_03] Which brings us to the second bonus hustle, the Hermes agent workflow. This one honestly feels like science fiction to me. [SPEAKER_02] This workflow is purely about automating your freelance client acquisition. You use the Hermes agent alongside Claude Fable 5.2. [SPEAKER_03] Let's unpack the timeline on this. It costs about $3 in compute. And it requires maybe nine minutes of actual human time. [SPEAKER_02] The agent automatically searches Upwork for highly relevant freelance jobs. It analyzes the job description, builds the required sample project, and writes the customized proposal. [SPEAKER_03] It is kind of like having a hyperactive intern. One who scours the job boards, writes the cover letter, builds a custom portfolio piece, and just leaves it on your desk. [SPEAKER_02] That's a great analogy. It's a completely automated lead generation and pitching machine running in the background. [SPEAKER_03] Wait, doesn't an automated agent run the massive risk of sending garbage proposals to high-end clients? [SPEAKER_02] No, because the agent only prepares the draft materials. You still manually review, tweak, and approve absolutely everything before it officially submits. [SPEAKER_03] The agent does the heavy lifting, but you remain the final quality control filter. [SPEAKER_02] Exactly. You act as the editor, not the writer. You're simply managing your own tireless AI workforce. [SPEAKER_03] So what does this all mean? We've covered five distinct paths today, from 3D luxury sites to automated Upwork interns. [SPEAKER_02] The overarching theme is pretty clear across all these examples. The true value isn't just knowing how to prompt Claude Fable 5.2. [SPEAKER_03] The real advantage is using AI to affordably build customized working demos, like that interactive 3D perfume bottle or that playable bubble game ad. [SPEAKER_02] You prove your undeniable value to clients before they even hire you. You completely eliminate their financial risk by showing them the result up front. [SPEAKER_03] You show them the results. You don't just tell them about your coding skills. [SPEAKER_02] This raises an important question about taking action. Here's my advice for you listening right now. Review these ideas. Pick the one that closely matches what you can confidently deliver. Build a demo for a potential client. You are literally one project away from your first AI side hustle. [SPEAKER_03] We just discussed using AI agents to find jobs and write proposals for human clients. But, hmm, how long until the clients start using their own AI agents to evaluate our AI's proposals? Are we heading toward an economy where side hustles are just our AIs negotiating with their AIs while we sit back and watch?