There’s a lot that’s been written and said about Claude Fable. Anthropic’s most capable model is built to solve large-scale problems and tackle things that would otherwise take hours in minutes. It’s also had a bit of a rough start, by all accounts: it got pulled, then access was reinstated. That said, even now, users only have a couple of days to test it out before Anthropic switches over to credits-based use for the model. Clearly, this resource-intensive model is just too expensive to be offered to everyone.
Which makes this the perfect opportunity to capitalize on access. While long-term developers and coders are already pushing Fable to its limits with advanced projects, even beginners can benefit from having access to just a bit more, well, a lot more compute to throw problems at. The goal is to throw real-world problems at it, and even nudge you in the right direction with suggestions to improve on projects that you might be working on. Here are three beginner-friendly ways to benefit from using Fable.
Let Fable run instead of steering it
Take advantage of that million token context window
The biggest advantage of Fable is its ability to work with large codebases without losing context halfway through, and a significant boost in computational ability. Recently, I’ve been experimenting with dashboards, nothing that’s good enough for production, but something that can pull social stats into a weekly report. I wanted to expand that further to handle more data types. I’ve previously dabbled in Opus, and while it has certainly made headway, more often than not, I run out of usage limits and have to wait. Of course, in any other scenario, I’d tackle the codebase piecemeal, but if you’ve got a window of opportunity to brute-force the problem, why not have some fun with it? This is the kind of job that Fable should be able to handle much better, without drifting out of scope.
With Fable, you can scope out the job with explicit instructions about the project you’re working on, what you need it to do, and what you don’t want it to do, and step off for dinner. This is where Fable’s ability to hold a multi-hour job together actually comes in handy. With a million-token context window, it can practically hold an entire codebase for hours into a job, which helps prevent it from drifting. This allows it to accurately add in new data types, or refactor a spaghetti codebase, but also lets it make reasonable judgment calls on how to handle situations that Sonet or Opus would struggle with. Of course, all while also explaining its thought process. Compared to Opus, where I’ve invariably had to make multiple attempts to pull something off, Fable has reliably one-shotted simpler projects like the aforementioned dashboard.
So, if you’ve been wanting to pull together scattered stats, spreadsheet data, or feeds into a single dashboard, this is a great time to let Fable take a swing at it. As long as you’re explicit with your instructions, you should be able to get the results you need.
Let Fable make sense of the mess
Summarise and visualize your data across files and folders
While a lot of online chatter centers on coding, not every project has to be about coding. You can use Fable to finally make sense of the pile of files or documents sitting in your downloads folder. For example, I have a folder full of invoices that I want to chart out. I could, of course, open them one by one, extract the data, and put them into another Excel sheet. But this kind of repetitive work is exactly what AI was meant for.
Fable easily reads the data and creates a graph that tallies my incoming and outgoing cash flow in a visual graph that makes it extremely easy to see where my money is coming from and where it is going. While invoices are just one use case, you can use Fable to make sense of any form of data. Be it survey answers, research documentation, or even to outline a project report based on scattered documentation. If you’ve got a drawer or downloads folder full of statements, receipts, or reports you’ve been procrastinating on, this is a good time to hand over a full batch to Fable for easy summarization. Can Opus pull this off? Yes. Where Fable excels is that it increases the volume of documents it can handle by an order of magnitude.
Let it build an entire project from a rough idea
Let Fable’s agents handle the heavy lifting while you describe what you want
This last project is the one that I’d nudge an actual beginner towards first, simply because it needs the least input. It could be a simple personal tool like an expense tracker, or one to capture health statistics. My recommendation is to be as verbose as possible about scope and feature set, while avoiding technical jargon. If you’ve been curious about fleshing out an app idea, this is your opportunity to make it happen.
I’ve already talked about Fable’s larger context window above, but its ability to break a job into sub-agents comes in clutch for a task like this. It can send each new data source to be handled in its own thread in parallel, rather than working through them one at a time. That, predictably, speeds up processing dramatically.
Take advantage of Fable while you still can
Look, Fable is an extremely cost-prohibitive model meant specifically for people working on complex problems. As such, it is not a model that most people would need or be willing to pay for. That makes this free opportunity to test it all the more worthwhile. Moreover, it’s a great opportunity to tackle some of the tougher tasks you’ve been working on. Between its large context window and its ability to tackle multi-step tasks with sub-agents, it is surprisingly useful for solving problems that smaller models struggle with. Whether that’s building a simple app, organizing files and folders, or letting it chew through a large project, I’d encourage everyone to take advantage of it while you still can.
