06:22Nick Saraev · Tutorial
How to Use Fable In Every Aspect Of Your Life
A six-minute walkthrough of a screen-watching AI agent that logs your habits, waits for a pattern to repeat three times, then hands you the fix.
July 9thThe argument in one line.
Giving an AI agent one open-ended, long-horizon brief instead of a series of one-off prompts lets it build its own generation infrastructure, turning ideation itself into the bottleneck it removes rather than the deliverable it produces.
Read if. Skip if.
READ IF YOU ARE…
- You build with AI coding agents (Codex, Claude Code, etc.) and want a concrete pattern for unattended, multi-hour agent runs.
- You work in e-commerce, dropshipping, or paid ads and want a faster way to generate and visually validate product concepts before committing money to them.
- You're curious how MCP servers let an agent call third-party generation tools (image, video, hosting) directly instead of through a manual UI.
SKIP IF…
- You're looking for a vetted, profitable product to sell — this is a concept-generation demo, not market validation or real sales data.
- You want a no-cost workflow — this requires paid subscriptions to at least three platforms (ChatGPT Plus/Pro, Higgsfield credits, GPT Image credits).
- You're not comfortable giving an agent real API tokens and shell access with minimal supervision.
The full version, fast.
Rather than prompting an AI model for one deliverable at a time, the video argues you get far more value by prompting it to build the infrastructure for generating thousands of candidates, then applying human judgment to pick winners. Using GPT-5.6 Sol through Codex CLI with a single markdown brief, the agent autonomously ideates 100 fictional products per batch, shoots packshot and lifestyle stills with GPT Image 2, animates two video ad spots per product via Higgsfield's MCP server, runs a self-QA loop on every asset, and publishes the results to a Netlify-hosted showcase site — all without further human prompting. The reusable pattern is the brief itself: state the mission, tell the agent the human is away and it must decide and proceed, define a self-critique loop, set hard non-negotiable guardrails (no real spending, fictional brands only), and parallelize many worker instances against a shared manifest.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 01:03
01 · Cold open
"Twenty products. Now in motion." title cards over a montage of generated product videos, setting up the reveal.
01:03 – 03:00
02 · Touring the 80+ generated products
Scrolls the auto-published showcase site: Rillway (reversible rug), Aeromere (incense cones), Waxora (wax tiles), Cabloon, RisePlane, Orbit Loop, Beam Vora, Twist Fall and more.
03:00 – 05:03
03 · The core idea: let the model prompt itself
Explains the philosophy: don't ask the AI for one deliverable, ask it to build infrastructure that generates millions of candidates, then apply human taste to filter.
05:03 – 05:28
04 · The 4-step pipeline
Diagram: Ideate → Stills → Video Ads → Publish.
05:28 – 06:10
05 · The platforms you need
Diagram: GPT-5.6 Sol (orchestration via Codex CLI), GPT Image 2 (photography), Higgsfield (video ads), Netlify (hosting).
06:10 – 08:13
06 · Setup: Codex CLI + GPT-5.6 Sol
ChatGPT signup/upgrade, installing Codex CLI, switching the active model to GPT-5.6 Sol at medium reasoning.
08:13 – 11:28
07 · Setup: Higgsfield MCP
Signs up for Higgsfield, then connects its MCP server to Codex so the agent can call image/video generation tools directly instead of the manual UI.
11:28 – 12:30
08 · Setup: Netlify hosting
Creates a Netlify personal access token and hands it to the agent so it can deploy the showcase site itself.
12:30 – 14:39
09 · The master prompt, line by line
Reads through PIPELINE-PROMPT.md: mission statement, "human is away, decide and proceed," the per-batch pipeline, a self-QA loop, and hard non-negotiable guardrails.
14:39 – 15:35
10 · Parallelize everything
Instructs the agent to spawn many parallel workers (~20 at a time, 5 products each) coordinated through a shared manifest with a lock.
15:35 – 16:20
11 · Launching the engine
Pastes the markdown brief into the terminal and starts the run.
16:20 – 19:05
12 · Watching it generate live
Tours the Higgsfield dashboard mid-run: simultaneous video generations across many product concepts, with occasional visible errors the self-QA loop is meant to catch.
19:05 – 20:12
13 · Get everything free
The full prompt set is free at "Maker Zero"; pitches "Maker School," a paid AI-automation community with a 90-day first-paying-customer guarantee.
Lines worth screenshotting.
- Ideation, not execution, is the task AI agents are currently better at than humans — use them to generate the option set, then apply human taste to filter it.
- A single long-horizon brief that tells an agent 'the human is away, decide and proceed' produces a fundamentally different (and more autonomous) result than a series of one-off prompts.
- Grounding video generation in a still image first produces more coherent output than generating video directly from a text prompt.
- An MCP server turns a manual generation UI into something an agent can call directly, which is what makes unattended multi-tool pipelines possible.
- A self-QA loop — generate, critique your own output in writing, regenerate, repeat up to three times — is what catches obviously broken results (like a key not seated correctly on a lock) without a human reviewing every asset.
- Parallelizing many agent workers against a shared manifest with a lock turns a linear five-to-ten-minutes-per-product process into a batch that finishes in about the same time as doing just one.
- Non-negotiable guardrails (no real purchases, no billing access, fictional brands only, publish only to the sandboxed destination) are what make it safe to tell an agent to run unsupervised for hours.
- A brief phrased as 'build the infrastructure to generate millions of these' produces broader, more useful output than a brief asking for one specific deliverable.
Prompt the agent to build a system, not a single deliverable.
WHAT TO LEARN
The gain here doesn't come from a better model — it comes from restructuring the brief so the agent generates its own options at scale and a human only has to pick winners.
01Cold open
- A single unattended agent run can plausibly produce dozens of finished creative assets, which reframes what "one person's output" can look like.
02Touring the 80+ generated products
- Volume alone isn't proof of quality — judging a generated catalog requires looking at more than the best few highlighted examples.
03The core idea: let the model prompt itself
- Reframe the request from "make me X" to "build the infrastructure that generates many versions of X" — that shift is what unlocks scale.
- Human judgment is most valuable at the filtering stage, not the generation stage, once volume is cheap.
04The 4-step pipeline
- Breaking a creative process into discrete stages (ideate, produce, animate, publish) makes each stage independently automatable.
05The platforms you need
- A workflow like this depends on stacking several narrow best-in-class tools rather than one all-in-one platform.
06Setup: Codex CLI + GPT-5.6 Sol
- Model choice and reasoning-effort level are levers worth tuning deliberately — medium reasoning was chosen here specifically to balance quality against cost.
07Setup: Higgsfield MCP
- An MCP connection is what lets an agent call a creative tool's functions directly instead of a human operating its UI — that's the difference between assisted and autonomous.
08Setup: Netlify hosting
- Giving an agent a scoped access token (not full account access) is what makes it reasonable to let it publish on your behalf.
09The master prompt, line by line
- Explicit non-negotiable guardrails in the brief itself (no real spending, fictional only, sandboxed publishing) are what make unsupervised operation defensible, not just convenient.
- A written self-critique step, capped at a few rounds, is a low-cost way to raise output quality without human review of every asset.
10Parallelize everything
- Coordinating many workers through one shared, locked manifest avoids duplicate or conflicting work without needing a human traffic cop.
11Launching the engine
- A well-designed brief can be reused as-is across runs — the setup cost is paid once, not per batch.
12Watching it generate live
- Even a well-instructed agent produces visibly broken output some of the time (e.g. physically implausible product interactions) — plan for a review pass, not zero defects.
13Get everything free
- Giving away the full technical workflow for free, then charging only for accountability/monetization support, is a distinct offer from charging for the technique itself.
Terms worth knowing.
- MCP (Model Context Protocol)
- A standard that lets an AI agent call a third-party service's tools (like an image or video generator) directly and programmatically, instead of a human operating that service's web UI.
- Codex CLI
- A terminal-based coding agent interface that lets a model like GPT-5.6 Sol read, edit, and run code and commands directly in a project directory.
- Long-horizon creative session
- An agent run framed around an open-ended mission rather than a single request, intended to continue for many cycles without the human answering follow-up questions.
- Self-QA loop
- A cycle where the agent generates an asset, evaluates its own output against a rubric, and regenerates it (up to a capped number of rounds) before considering it finished.
- Bypass permissions / YOLO mode
- An agent operating mode that skips per-action confirmation prompts, letting it act autonomously at the risk of taking an unwanted action.
Things they pointed at.
Lines you could clip.
05:17
“You have to let the model prompt itself.”
03:44
“Ideation is the core skill that agents have that human beings currently do not have.”
02:56
“This is like reality shipping. We are instantiating these things in reality from scratch.”
12:41
“The human is away. You will not receive answers to their questions. Do not stop to ask. Decide and proceed.”
15:18
“You can legitimately run this entire thing in like ten minutes and make a million ads.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphor
00:00So to test the business potential of GPT 5.6 Soul, I had it run essentially a fully autonomous product creation and creative generation loop. I had it come up with the products themselves, create high quality image advertisements using GPT image two, high quality videos using Cdance and Cling models, and then weave that together into fully autonomous marketing campaigns.
00:22And I did this because I don't think that just looking at benchmarks is sufficient enough to really know whether or not one model is better than the other. To really understand, I think you need to test it to real world capability.
00:34That is how good it is at designing things that realistically people would would be interested in buying. So the end result is over 80 high quality products. And what I wanna do in this video is I actually wanna give you guys everything that you need to recreate this yourselves, including all the prompts and so on and so forth.
00:48I'm also gonna give you this website over here, which contains everything that I fed into the model in order to generate these, so that you guys can also set up your own, like, autonomous Shopify stores. Totally autonomous, you know, Facebook ads pipelines to validate ideas, and so on and so forth. This is Rilway, a beautiful reversible hallway run.
01:06You can see it has kind of like a design on the other side, and that design is kind of nature inspired. It even went through and then shot a video of what that might look like in a house. This over here is Arrow Mirror, which is a low smoke incest cone blended for a queen cedar and rice paper scent.
01:21And you can see here that the whole idea is this is like kind of a conical incense thing that you put in this beautiful bowl and it looks super aesthetic, but it also, you know, seems really relaxing and chill. How about Waxora, which is snap apart wax tiles that pair two complementary scents in clearly divided colors?
01:38Obviously, it's going for a vibe here. Right? But have you seen products like this?
01:41These are marketed extraordinarily sexyly. You know, there's like AirPods carrying cases.
01:46There's decks and and ports and hubs for devices. I really like this one, Cabloon, which is a tiny magnetic desk guide that routes charging cables without any adhesive clips.
01:56This looks really simple and straightforward and based off the minimal aesthetic vibe, I'd a 100% have this on my desk. How about RisePlane, which is like this laptop stand, which significantly improves the ergonomic effectiveness of using, you know, one of these MacBooks.
02:10Or maybe Orbit Loop, which doubles as a finger hold for iPhones, but also allows you to stand the phone up. Beam Vora, which is a slim monitor mounted webcam that creates broad eye level illumination. And Twist Fall, which is a inverted ceramic spice mill.
02:25I mean, I could go on and on and on. But the actual creation of products is now capable of being completely automated. And I can imagine with a little bit of work, what you could do is you could build a pipeline that does things like scrape Reddit threads or customer support channels and stuff like that with features that you want, products that people desire in the market, and then essentially fully autonomously generate these, run ads for them, validate them, and then send some sort of schematic or diagram to a person and or, like, a machine shop to actually have this stuff generated and shipped.
02:56You guys know drop shipping? This is like . Like, we are instantiating these things in reality from scratch, which is super cool.
03:03You know, so how did I actually do all of this? The important thing that I wanna make super clear is I'm no longer feeding in a specific prompt. I'm no longer actually giving it, like, a one to one pipeline.
03:14What you have to do, in this case, the GPT 5.6 sold, it really can apply this to anything. So you have to let the model prompt itself. Instead of saying, hey, man.
03:22I want you to create this cool advertisement for, like, some headphone product. You have to say, I want you to build the infrastructure that would enable you to generate millions of these, and then I want you to run them on autonomous loop with some form of self verification.
03:35Here are access to all the platforms that you can use Go. And then what you get is you get this massive library of candidate options that you get to pick. What you'll quickly realize when you start doing this at scale is that ideation is the core skill that agents have that human beings currently do not have.
03:51And that's what you should get AI to do. Rather than try and have AI, you know, do the actual deliverable right now, Because of its taste, because of things like stereotypical AI speech, and so on and so forth, it's far better and more effective to have AI help you with the top of the funnel, aka the generation of a million billion ideas, and then narrow that down using your own human intelligence to some candidate options to pick.
04:16And so all of this really, if you think about it, is in service of that end. It is getting AI agents to come up with both the product features, but also, like, some sketch images and concept images of what it could look like. It's getting AI to feed this, let's say, into some sort of advertising network.
04:32These are all candidate options, so maybe you spin up 20. And then what happens is human taste filters them out and then picks ones. You can, as mentioned, entirely automate this process, and I have here.
04:41But, you know, I think just not even for the specific use case, just in general, if you take into account that human beings are very slow and inefficient at coming up with ideas, you know, our brains are naturally wired to iterate over all known patterns. Whereas AI agents are inherently extremely fast, and this is kind of what they're built for, iterating over all known patterns.
05:00You can make some really, really cool stuff out. So we're gonna talk tools in a sec. Right now, the pipeline looks kinda like this.
05:06AI will ideate. You will generate stills. So these are some form of images.
05:09I find if you don't ground it in images, which are currently higher quality than videos, typically, videos are nonsensical and shit just looks kinda wild. Then you do some form of video ads. So in this case, I did two spots per product.
05:22And then finally, you publish some sort of auto deployed site, you know, if you really did wanna scale this up to some sort of automated advertising system. The platforms that you need to make something like this happen is you obviously need the text model, which in this case is GPT 5.6 SOL. Then you need some sort of image model.
05:37And in this case, I use GPT image too because I think it's it's just easily the best. After that, you need some form of image or video ad generation model. And I'm accessing all these through Higgs field.
05:47The value there is they have this MCP, and I'm gonna run you guys through how to use it all in a sec, which means you can just give GPT 5.6 sole the server and just have it come up with all of the creative decisions. And then finally, you know, I'd recommend you have some way to host things. I And really like Netlify because I'm not affiliated with them whatsoever.
06:05But I really like Netlify because they like to instantly host things, deploy them to the back end. It's just super quick. Okay.
06:10So let's actually set this puppy up. As mentioned, we had four platforms. The first was ChatGPT.
06:15So just head over to chatgpt.com. You can then sign up for free over here. In order to use GPT 5.6 SOL, which is the newest model, you will have to spend a little bit of money.
06:24After you make your account, it'll look something like this. And then in the top right hand corner, you can upgrade. So in my case, I'm gonna go to personal, and I'm gonna go to plus.
06:31I'm Canadian, so this is a non freedom dollars. But if you guys are Americans, this will obviously be in a different right suit. Once you set this up, I highly recommend we use Codec CLI.
06:40Right now, to me, this is just the easiest and most straightforward way to use GPT 5.6 SOL. And it's fairly easy. All you really have to do is just open up a terminal app.
06:49So, like, I use this one called Ghost TTY. And then I can open up Codex in any one of these instances. So you could see here, you know, I have this, I have this.
06:57These are all just different instances that are running. On the top half of my screen are a bunch of Fable instances, and then on the bottom half are a bunch of Codec instances. And they're pretty cool.
07:05Although you'll see that it's auto selecting 5.5 pipe, and I'll show you 5.6. So all you need to do in order to install it is just head down here to where it says get started with Codec CLI. And then I'm just on Mac OS or Linux, so I'm just gonna copy this.
07:17I'm gonna go back to one of my terminals, which again, you know, you can open up a terminal. You can open up ghost t t y. These are all just different ways to access the same app.
07:24And then just paste in this command. This will then install codecs for you. And then once you're done with the installation, you're good to go.
07:29Now, I can actually type in codecs. And I personally really like using what's called the yellow mode. So I go slash slash yellow.
07:36Don't get intimidated by all this code type stuff. There really is nothing super intimidating. After you're done, then you just go slash model.
07:42And then what we wanna do is we wanna use GPT 5.6 So I'm just gonna move over to that. And I recommend medium reasoning for okay levels of quality without it absolutely skewing your cost. Okay.
07:51So now we've set up GPT 5.6 SOL. Because you've also set up, like, your ChatGPT subscription, you will have access to images built in. I will note that these images will eventually run out of credits, and you may have to purchase more credits.
08:03But because the interface changes so often, I don't wanna, like, make you hard code it. So I would just look up, like, GPT image credits or buy GPT image credits, something along those lines, and you guys will get a good link.
08:12Okay. Now I'm gonna set up Higgs field, which, as mentioned, is extremely stimulating when you make it on the page because they're just running all these videos simultaneously. So your computer may heat up a little bit like mine does, but, basically, this is just like a video image and then audio library.
08:27And, you know, when now that we're in an environment where there's, like, so many thousands of different models you could choose from, the value proposition of these sorts of aggregators is they just aggregate them all into one place, and then they give you a layer where you can just, like, the API and the same credentials just call any one of these models.
08:44So that way, of, you know, signing up to Gemini, but also to Cdance and blah blah blah, just do it all internally. And you you don't have to use these guys. This is just very straightforward.
08:51It's one of the simplest ways that I find actually whipping this up. So eventually, you're gonna want is you're gonna want the video model, but obviously, first, we need to sign up. So I'm just gonna open this in an incognito tab, and then in the top right hand corner, try exit out of this cookie notification.
09:03I'm just gonna click sign up. And once I click sign up, this little model's gonna pop up. I think they're giving you some additional credits right now, either for free or, I don't know, some very low cost.
09:13So I'm gonna continue with Google on this case, I'm just gonna sign up. Okay. Now, we can use these models really easily just by going to this little video tab and then typing whatever the heck we want.
09:22And as you can see, we've actually generated the videos themselves in this UX already. I mean, that's what's going on with one of these, I don't know, QR code phone cases window ink. I guess it's a wallet case.
09:31But this is pretty inefficient because if you think about it, then you have to, like, actually manually create your prompt on the left hand side of your time. You have to wait for the outputs and stuff like that.
09:40A much more effective way to do this is to use what's called the MCP, model context protocol, which is the server that Higgs field makes available. Basically, a little API that allows you to have AI agents do all this stuff autonomously instead.
09:54And this is gonna be the foundational hotbed upon which we build all the rest of this value. So to use this, all you have to do is click on the MCP and CLI up here. It'll say Higgs field MCP for any AI.
10:04Go to ChatGPT, in our case, simply because that's what we're using. And then you can turn on developer mode, create the Higgs field app, and get it. If you're using the Codec CLI, actually, you can actually just, like, ask it, hey, can you use Higgs field for chat GPT?
10:16So this is what I'm gonna do. I'm just gonna go copy, and I'm gonna go back into my Codec instance, which I have been in ghost TTY, and I'm gonna paste. I'll say, can you set this up?
10:25I'm also gonna give it a smiley face because when they do eventually turn me into paper clips, I want them to do it softly. This will eventually open up this little codex request here.
10:34So I'll then click allow, and then it'll say authorization complete.
10:39You may close this window. So let me go back over here, and then you can see it's now verifying the Codex permissions and sort of signing up and stuff like that. And now, you know, after it's done the verification of MCP tools, we're basically good to go.
10:51And now that it's set up globally in Codex, you can do whatever the heck you want. Worth noting that you will have to reset the Codex chat in order to load its tools.
11:00And then what you do in order to check whether or not you have it is you just go first of all, to starting MCP servers two of three Higgs field, which is good. I'm just gonna go slash m c p.
11:11It'll load up your MCP inventory, and they can actually see. So we do now have scroll up here.
11:17Exit out that little Chrome extension. We have Higgs field with auth, o auth, tools, animation, actions, balance, capture, whatever. Like, these are all of our these are all of our tools.
11:26Okay. Now the only thing you need is you need a website hosting platform. So in that case, I'm gonna use Netlify, as mentioned.
11:30You guys can use whatever you want as well. Netlify is pretty easy. Just head over to Netlify, and then you can go push your IDs to the web.
11:38Then just sign up over here. And, you know, in my case, I'm just gonna sign up with Google. And so you have it, just head over to I think it's like Netlify Labs, probably.
11:45No. Sorry. It's user settings.
11:47Go applications, and then what you want is you want a personal access token. So you can then go new access token, and I'll say, I don't know, GPT. Just set it to never expire down here.
11:57Generate the token. And then what you wanna do is you just wanna copy that and then feed that into your codecs. So go back over here, and then say, you know, add this to the workspace.
12:09It is my Netlify token. You will need it for something.
12:14K. Cool. It'll probably yell at you and say you're publishing your API details, you know, locally, and that's okay.
12:21I've actually already given this the one that I gave to Cloud Code, so I'm just gonna delete this so that you guys don't have full access over my websites. Okay. And then you can see it's stored in this Netlify auth token in users, next.
12:30Right? This is. Right.
12:31So what does the actual comp look like? It's called the Autonomous Product Ad Engine. You're running an unattended long horizon creative session.
12:38You have full access to the Higgs Field MCT. So these are all generation tools. Sorry.
12:43Let's go back here. All generation tools, the supercomputer scope workflows. You just see flow product come up.
12:47And And then here's, like, the the real crux of it. The human is away. You will not receive answers to their questions.
12:54Do not stop to ask. Decide and proceed. So long horizon creative session, the human is away.
13:00These are two phrases I've been using quite often over the course of last week when working these models, not just GBD 5.6 SOL, but also Fable five. And the reason for this is because if the human is away, the model needs to continue going.
13:11And so rather than wait to ask you, it will just pick the highest probability thing if things will succeed and then just continue with that. Obviously, there's some caveats, and this could be kind of sketchy. If have full YOLO mode or, like, bypass permissions mode with their clad family of models, you know, this could eventually lead you down some path that you don't want to.
13:29And maybe, I don't know, it deletes your entire hard drive and bills you a billion dollars. But probability of that is quite low. I think it also run some sort of sandbox container if you wanted to.
13:38Okay. So, yeah, the mission is to run a twenty four seven marketing production engine for simple physical products. You know, I gave it some examples here, but I wanted to just come up with whatever the heck it wants.
13:49So skincare, footwear, apparel, hardware, drinkware, home goods, all fictional brands you're gonna venture to solve. Treat this as your chance to apply your full creative potential. The output of the session will be shown to 500,000 people as a demonstration of what a French model does with total creative freedom and unlimited generation budget.
14:02Now I say unlimited here because I have lots of money, and I like converting the money I have into, you know, interests on YouTube, Instagram, and these other platforms. Obviously, if you guys have a very, very limited budget, you should not use the term unlimited generation budget. Here's the pipeline per batch, and it just repeats until told to stop.
14:19It'll start by ideating a 100 products, then it'll shoot them with GPT image two, and it will animate them with Higgs field to get the videos, and then it will publish. And finally, for quality, we need to make it check its own work. So every asset will go through a self QA loop before it ships.
14:33They'll generate, look at the results with its own vision, critique it in writing, regenerate it, and so on and so forth. And this is a really cool thing, parallelize. What's really cool is if you try to do this like the old school way, which is sort of like the linear way, like, I come up with a product, I verify the product is okay, I shoot that, I turn that into a video, and then I publish it on the website.
14:53That's gonna take, like, five to ten minutes per run. And then if you wanna do another one, it's another five to ten minutes. What we do instead is we parallelize it.
15:00Basically, we just do all of that simultaneously. So we come up with a 100 ideas. For every idea, we come up with a 100 or 200 images.
15:06For every image, we come up with, you know, a 100 or 200 videos. And we do all that in the same amount of time as it would've taken just to do one. Because Codex and the GPT 5.6 Soul make use of sub agents really, really effectively.
15:18You can legitimately run this entire thing in like ten minutes and make a million ads, you know, for whatever uses that you're going to be using for the next like three months. That's probably not necessarily a great idea because I bet you the model tell just will get even better by then and it can be even cooler things.
15:32But just wanted to give you guys a quick example. Okay. So what I'm gonna do is I'm just going to grab the like markdown file here and I'm just gonna copy and paste this into my ghost TTY terminal.
15:40Paste that in. And then, now, it's just going to run. And the whole idea here is, you know, this sets a goal for itself based off of this, and then it will run.
15:50And the reason why I always like having this set up in, like, a six panel way is just because this allows me to do other things while it is running. So this will this will set everything up. This will hook to the website.
16:01This will publish the website. It'll even pick an idea. It'll just serve you the link, basically, at the end, which is really sweet.
16:07Obviously, as mentioned, this is going to consume Higgs field credits, which are gonna cost. This is gonna consume GPT image credits, which are gonna cost, and this is also for consumer usage if you are on the the chat plan.
16:17But, you know, AI is free these days, and that's kinda how it works. So while we're at it, let me show you the outputs that we're getting inside of Higgs Field. I'm just gonna go back to this little video panel, and then we're gonna take a look at what's going on.
16:27So as you can see here, this is a five second premium product beauty shot. Lock the exact bottle geometry, ivory pump, glass reflections, label design, and a correctly spelled word, clear hour, k, from this frame.
16:39And you can see essentially what it's doing here is it's already generated an image of this beautiful spray bottle called clear hour, and now it's generating a video for that. If I scroll down, I mean, we're doing a lot with right now.
16:50Like, we're we're publishing all eight of these simultaneously, which is pretty cool. Before, we were using Nest Lock, which is like your little I don't know.
16:58It's like a key sort of binders thing. As you could see, we're kinda sticking it up. I I mean, you know, would I consider this to be perfect?
17:03Like, no. You're gonna get some inputs that don't really make sense. Like, the key isn't going all in the way.
17:08But that's what the verification loop is for. It tends to catch most of these. This is a man wearing refined vertical travel menswear.
17:15I mean, this looks pretty sweet. Right? Like, it's pretty sexy.
17:18Damn, I wish I had those pants right now. I'm wearing some five old Levi's. You can see that we generated multiple variants of this sort of nest lock thing, which is kinda neat.
17:27It shows you like the range of motion, moves around. This is interesting. It's called pom turn.
17:31It's basically like a screwdriver that I I suppose ratchets really easily. Maybe it's just like a clean sort of ratchet. We even have videos for those pants as shown over here.
17:42It looks like it's a slow rotation, which is neat. I'm a little bit laggy because obviously, I'm doing a fair amount right now. Rain roll.
17:49It's like one of these bucket hats except a bucket hat that is 100% rain resistant. Man doesn't look too happy about it, but what are you gonna do?
17:56And then I mean, we just have so many of these. What's worth pointing out is because, as mentioned, there's a variety of these different models in Higgs field, like, the MCP connector is being used across a ton of different approaches.
18:07Like, is Clang three point o turbo. But if you scroll up, you'll see that we're also using c dance too. So we're actually using, like, a ton of different models and different approaches.
18:14And this is what I mean by just, like, letting the model do it on its own. How about this? This is like a makeup product for, like, vertical ads that it just came up with.
18:21The reason why it's laggy again is just because I have multiple of these going simultaneously. That looks clean as hell. I love the font.
18:27I love the design. I love the way that it's kinda like, you know, a little padding. Yeah.
18:31Yeah. And, you know, we just have this running completely autonomous here right now. There's no human in the loop.
18:37I am just straight publishing these to a website, And, hopefully, you guys see how easy it would be to weave this into maybe some sort of, like, ads based workflow. Or in addition to that, we, you know, publish this on the Internet. Uh, I don't know.
18:48Make content about it. Do anything like that. You guys are probably gonna start seeing a lot of these ads in the near term.
18:53So be careful with that, obviously. Make sure the products that you buy are genuinely real products. But, yeah, the future advertisers of the future are using this exact workflow right now.
19:02They're just not making YouTube videos telling everybody that. Okay. So hopefully you guys appreciated that video.
19:07You guys saw how cool and straightforward and easy it is to do. What I want you guys to do right now is I have all of the prompts and everything that I used to create this down below in Maker Zero. It's a free community that put up that basically just hosts all of the resources that I give out because I don't believe that actual technical AI automation knowledge should be gated.
19:24I'm happy to just give all that away for free. If you guys like this sort of thing and you want to convert this big product generator into some sort of monetizable service or product and sell it to people, then check out Maker School. That's gonna be the first link down below in the description, which is my AI automation community where I literally walk you through the monetization.
19:43So that involves, like, ninety day accountability, a guarantee where you'll get your very first paying customer in that time period, or, you know, you get a full refund. Includes a big group of people that are rooting for you, cheering for you, and showing you what's possible.
19:54And we have so many people closing big deals for systems like I just showed you every day. It's not even funny. Probably the community with the highest actual, like, return on investment rate in school right now.
20:04So I'll stop tooting my own horn. Thank you very much for your time, and I really appreciate every second that you've been watching this video. Looking forward to catching all y'all in the next one.
20:11See you.
The Hook
The bait, then the rug-pull.
To find out whether a frontier model is actually good at real-world business tasks — not just benchmarks — the creator gave an AI coding agent full tool access and one open-ended brief, then stepped away and let it invent, shoot, animate, and publish an entire catalog of fictional products on its own.
Named ideas worth stealing.
05:03list
The 4-Step Pipeline
- Ideate (100 fictional products per batch)
- Shoot (GPT Image 2 packshot + lifestyle stills)
- Animate (Higgsfield, two video spots per product)
- Publish (auto-deployed showcase site)
The repeatable per-batch loop the agent runs until told to stop.
Steal forany unattended multi-asset content-generation pipeline
12:30concept
Master-Prompt Guardrails
- Human is away — decide and proceed, don't wait for answers
- Self-QA loop: generate → critique in writing → regenerate, max 3 rounds
- Parallelize via a shared manifest with a lock
- Hard non-negotiable rules: no real purchases/billing, fictional brands only, publish only to the sandboxed destination
The structural elements that make a long-horizon unattended agent run both productive and safe.
Steal forwriting briefs for any autonomous multi-hour agent session
How they asked for the click.
VERBAL ASK
19:05product
“the first link down below in the description... my AI automation community where I literally walk you through the monetization... you'll get your very first paying customer in that time period, or you get a full refund”
Soft-sells the paid community only after giving away the entire prompt set and workflow for free through a separate free community, framed as "for people who want to convert this into a monetizable service."
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
Visual structure at a glance.
