Claude Fable 5 Use Cases You Must Do NOW (Or Lose Thousands in 1 Week)
CChase AI
Computing/SoftwareSmall Business/StartupsVideo & Computer GamesInternet Technology
Transcript
00:00:00The most powerful AI model ever, Fable 5 is back, but we only have a week to play around
00:00:05with this thing before we lose it to API pricing. So the question you should be asking yourself is
00:00:10what projects can I use Fable 5 on over the next seven days to get the most bang for my buck?
00:00:17How can I actually squeeze every ounce of juice out of this model? Well, in this video,
00:00:22I'm going to help you out as I show you five different projects you can point Fable 5 at
00:00:27and get your money's worth. So with that, let's hop in. So the first and arguably the best use case for
00:00:33Fable 5 is simply cloning software that already exists out in the real world, probably software
00:00:38you pay for. Well, why are we paying for it? Why don't we just build a clone ourselves and customize
00:00:43it to our needs? Fable 5 is great at doing this. And in this demo, let's have it clone Whisperflow.
00:00:49Ton of you probably use Whisperflow or something like it. Well, why are we paying for this? Furthermore,
00:00:53why are we giving our data to someone like Whisperflow and having that information that we speak into our
00:00:58microphone go to some cloud server? Why don't we just create a Whisperflow that's purely local,
00:01:04runs on our machine, is faster. And again, we can customize to our heart's desire. Well,
00:01:09that's exactly what Fable 5 can do. Now, when we work on these big projects, something we do need to
00:01:15keep in mind is our usage limits. Up until July 7th, we can use Fable 5 with our max plans. However,
00:01:22we can only use up to 50% of the plan's weekly usage limits. So we need to be smart about how we do this.
00:01:27We need to be smart about the sort of prompts we create. It doesn't make a lot of sense for me to
00:01:32just say, hey, Fable 5, go recreate Whisperflow. We can probably do better than that. In fact,
00:01:37what we can probably do is we could use something like deep research, aka dynamic workflows inside of
00:01:43cloud code using a model like Opus 4.8, have it figure out a plan that actually makes sense.
00:01:50We could even use something like codex to check that plan. And once we have a plan that makes sense,
00:01:54we hand it off to Fable 5. We do some of the upfront grunt work with Opus and let Fable 5 handle the
00:02:01rest. So that's exactly what we're going to do. So again, I'm on Opus 4.8. I said forward slash deep
00:02:05research. I want to come up with a plan to clone Whisperflow. I want you to do some deep research on
00:02:10Whisperflow works, what we would need to recreate its base functionality on our computer. Furthermore,
00:02:14I'd like to recreate it locally. So I want it to be a local model running on a Loma that essentially
00:02:19does what Whisperflow does. So we're going to run this. It's going to come back with a plan.
00:02:24And if that plan makes sense to us, we're going to go ahead and hand it to Fable 5 and have it go to
00:02:28work. So it went ahead and finished the deep research and figured out, hey, this is what a
00:02:31Whisperflow clone should look like. Here's how it would work on your computer. And here's sort of the
00:02:35local architecture I'm thinking about. Now, what we want to do is we want to take this whole report
00:02:39and we want to turn this into a prompt we can hand to Fable 5. And we can have Opus do that just fine.
00:02:44Ideally, we set up this prompt so it makes sense if we do forward slash goal. Remember,
00:02:49forward slash goal is for long running agentic tasks, big projects, things that are perfect for
00:02:53Fable 5. And with forward slash goal, we're saying, hey, this is what we want to do. And here's
00:02:59sort of the success criteria. And it's just going to keep working and working and working
00:03:02until it gets to your end state. So perfect for things like this. So it went ahead and created
00:03:08that prompt for me. So I'm just going to go ahead and copy this thing. We're then going to head and
00:03:13just switch the model over to Fable 5. We pasted the prompt in there and we just let it go to work.
00:03:22And after some back and forth to get the visuals working, we got this, which is my version of
00:03:27Whisper Flow, but entirely local. Nothing leaves my computer. Doesn't have all the bells and whistles
00:03:32of Whisper Flow, but it does the basics. It listens to what's going on with my microphone.
00:03:37It transcribes it. It sends it down to the local AI model to clean it up. And when I'm done talking,
00:03:42it just populates it inside the text box. Let's see what it gives us.
00:03:49So, hey, nothing leaves the computer. It doesn't have all the bells and whistles, et cetera, et cetera,
00:03:52et cetera. So it just took everything I said and put it inside here. Now, all that being said,
00:03:59is Whisper Flow clone the craziest thing ever for Fable 5? No, it can actually do a lot more than that.
00:04:04But it just sort of depends on what you want to clone. I think you sort of just use the template
00:04:08I gave you, which is do some deep research, figure out how that particular app actually works,
00:04:13figure out how you want to customize it, get your prompt in order, and then bring it to Fable 5.
00:04:18I definitely do not suggest using dynamic workflows with Fable 5, or you're just going to burn through
00:04:23all of your usage. Now, let's move into use case number two, which is using Fable 5 to
00:04:27do a complete teardown and diagnosis of how you use Claude code and how you can improve. Now,
00:04:33I'm not talking about how you use Claude code in terms of usage. I'm saying we're going to have
00:04:37Fable 5 look across all your previous sessions, take a look at how you use Claude in terms of your
00:04:43skills, your automations, your tasks, and then figure out what you're doing right.
00:04:48What you're doing wrong. And more importantly, what we can do to improve this. Does this mean
00:04:52changing our skills, creating new skills, adding new automations? So this is essentially doing an
00:04:57audit of how you're using the tool itself. So here's a look at the prompt. We're saying reflect on our
00:05:02past Claude code sessions to find the highest leverage improvements to my setup. Use sub-agents to pull
00:05:07raw signals from the transcripts. You cluster them across sessions and decide per cluster whether it needs
00:05:12a new skill and automation, a fix or nothing. Write the candidates in this MD file. And we're saying,
00:05:19hey, at first it's just a diagnosis. I want to see what it comes back with before it executes
00:05:22anything. Now, if you're wondering how I'm coming up with these prompts and this prompt structure,
00:05:26this is coming from Anthropix official documentation when it comes to prompting Claude
00:05:30Fable 5, because there are some nuances between how you want to use Fable and Mythos versus something
00:05:35like Opus. So let's see what it comes back with. So Fable 5 ran through my last 39 sessions,
00:05:39and this is what it came up with. It broke it out into three different batches based on how much
00:05:44leverage you'd think it would give me. And they ranged from creating new skills to setting certain
00:05:49skills as automations and some simple changes to things like my Claude.md. So this is a really
00:05:55simple use case to improve your workflows within Claude code. And it's something you'll get a lot more
00:05:59out of if you fall into the power user side of the equation. So before we jump into the next use case,
00:06:06just want to give you a quick word from today's sponsor, which is me. I just released the Claude
00:06:10code masterclass not too long ago, and it is the number one way to go from zero to AI dev,
00:06:15especially if you don't come from a technical background. We update this every single week and
00:06:19all the resources you see in today's video, including my Claude OS can be found here inside
00:06:24Chase AI. Plus there's a link to that in the pin comment. So definitely check us out if you are
00:06:29trying to get more serious about your AI journey. Now let's go into use case number three, which is
00:06:33building your own agentic OS. What you see here is one I built with fable five, and this essentially
00:06:39acts as a custom wrapper over the top of Claude code. What we see here is the visual side of it,
00:06:45but what's most important is what's going on under the hood. And it's a perfect follow-on from use case
00:06:49number two, which is essentially codifying everything you do in your day to day, your week to week into
00:06:54skills and automations. But this gives us the additional advantages of certain visual metrics
00:06:59that we just can't get inside of the terminal. So for me, that includes things like content, right?
00:07:03What's been going on with my content game across multiple platforms. I can see things like my
00:07:08different like morning reports and things of that nature. This is all linked to Obsidian and I have
00:07:13all of my most used skills and automations over here on the right, which are just a click away.
00:07:18Now, again, I've done deep dives on this. I'm not going to turn this into a deep dive agentic OS video,
00:07:22but the most important part are all those skills and automations I'm talking about. You need to use
00:07:27fable five to come up with the skills and automations that make sense for you. For me,
00:07:31that has to do with like research content, things on my agency side, like sales and finance,
00:07:36all my individual tasks, fable five turns into skills and turns them into automations,
00:07:40if that makes sense. And in certain cases, we apply loop engineering to those skills as well. But that
00:07:46really depends on the use case. But this sort of customized agentic OS is pretty simple for fable
00:07:51five to create. And one of the best parts about this is that if you're in the agency game, you can
00:07:56package this since it's essentially a web app that anyone can put on top of their cloud code and sell
00:08:01it. Or you can clone it and give this to teammates who aren't going to use the CLI or aren't going to
00:08:07use the cloud app because they can add whatever they want to this. And you pretty much just wire
00:08:12up different skills and automations that you would use because it's just doing clod headless dash P
00:08:17under the hood, which luckily for us isn't pulling from API prices anymore since Anthropic walked that
00:08:23back a few weeks ago. Now use case number four comes directly from Anthropic itself, and that is code
00:08:28review and debugging. If you have a complicated project, if you have a huge code base, now is the time to take
00:08:32fable five and point it at that code base and see if you can figure out what code looks bad and what are
00:08:39actually bugs. And the problem for this doesn't have to be complicated. Hey, so I want you to take a
00:08:45look at this code base and I want you to do a full code review and also let me know about any bugs you
00:08:51find. And it's going to come back with whatever it finds. So after about five minutes, it found 45 raw
00:08:55findings from four parallel reviewers deduped down to 24. And then it took those 24 and it broke it down
00:09:01by severity. And it gives us sort of an explanation of what's wrong at each step, kind of like where the
00:09:07issue is. And then it gives us a specific fixed priority. And it's like, hey, do you want to go
00:09:12ahead and start working on these? Now, the cool thing about this is it found all these things wrong
00:09:15in a code base that isn't necessarily that complicated. There's things that are infinitely more
00:09:20complicated than what I'm doing here. So if you are in that camp of someone who has something very
00:09:24complicated, there is no reason you should not be pointing Fable 5 at it and at least having it get
00:09:29eyes on on the work you've already done. Now use case number five is what you see right here. It's
00:09:32having Fable 5 create whatever custom software you want, something that is going to require a long
00:09:37horizon. This is a video game built in a browser running on 3.js. And this looks wild. This is an
00:09:44insane accomplishment by Fable 5. You would not be able to create this using something like Opus 4.8,
00:09:50unless you knew exactly what you were doing. This is again, this is all running on the browser.
00:09:54This isn't like a downloaded video game. This is all browser graphics. And it's crazy. Now I wasn't
00:09:59the one who actually built that. This is an open source project from Bradfolk who created this using
00:10:04Fable 5 the first time it came out. But I think this is a great case study for the sort of things you can
00:10:09create. And the power here isn't just that Fable 5 built it. The power is that we can look
00:10:16at sort of how he created this from scratch. So the readme talks about the one document that they gave
00:10:23Fable 5 in order to create this. So the human partially wrote one document, which is this markdown
00:10:29file. And what is this? This is a PRD. This is a product requirements document spelling out,
00:10:34hey, here's what we want to build, right? The visual target is a current gen Unreal Engine 5 showcase
00:10:40footage. And then it goes through sort of like the pillars of this application, the instructions,
00:10:44the constraint, the floors, et cetera, et cetera. Like you said, this was only partially written by
00:10:49the human. So what you need to do, if you're trying to create your own sort of software or a game or
00:10:54whatever it is, some sort of crazy project that only Fable 5 can build, you need to nail this down.
00:10:58You need to nail down the PRD. Now Fable 5 can help you, but we want to be very conscious of our usage.
00:11:03So this is something Opus 4.8 can at least get it started with you, right? You should be able to
00:11:09create some sort of PRD with the specific instructions and with the specific requirements
00:11:14with Opus 4.8. Again, use something like deep research to help you do that and then bring it to
00:11:19Fable 5, essentially recreate what we did in the first use case. And after that, you pretty much just
00:11:23hand it to Fable 5 and you let it execute it across a long autonomous sessions, exactly like
00:11:29the forward slash goal scenario we did earlier. In this particular setup, Fable 5 wrote 21,000 lines
00:11:34of TypeScript across 90 plus commits to get what you just saw. So really cool stuff. And that's just a
00:11:39taste of what this model is able to build for you. So those are the five Fable 5 use cases you need to
00:11:45try out this week. As always, let me know what you thought about this video in the comments. Make sure to
00:11:50check out Chase AI Plus if you want to get your hands on the Cloud Code Masterclass or my exact
00:11:55Cloud OS setup. And besides that, I'll see you around.