JBS Weekly
Claude Fable 5 is included in your paid subscription right now, but that changes on July 7th when it moves to API pricing.
If you're a COO or ops director trying to figure out whether it's worth the cost, the answer depends almost entirely on whether you're prompting it correctly. Most people aren't.
The way you prompt Opus or ChatGPT will actively make Fable worse. This issue covers exactly what to change, with a concrete example of what a real Fable prompt looks like and a workflow for building one without having to think like a prompt engineer.
🛠️ This Week’s Build
The mistake most people make with Claude Fable 5 is treating it like a chat model. They send an instruction, get a response, tweak it, go back and forth. That works fine with Sonnet or Opus. With Fable, it makes the output worse.
Fable is built to receive a complete problem statement, not a conversation thread. The shift is this: stop giving it a route and start giving it a destination. That means context, goal, success criteria, and constraints, all in one prompt. Step-by-step instructions that help weaker models stay on track will trip Fable up.
Here is the format that works :
'I'm working on [larger task] for [who it's for]. They need [what the output enables]. With that in mind: [request].'
Five additions that make a real difference.
Tell it to plan before it acts: 'Plan this first. Ask me any questions. Then do it.'
Demand receipts: 'Before reporting progress, audit each claim against actual evidence. Only report work you can point to evidence for.'
Tell it whether you want a check or a fix: 'The deliverable is your assessment. Report findings and stop.'
Ask for a recommendation, not a survey: 'If you're weighing a choice, give a recommendation.'
And build in a fresh-eyes check: 'When you think you're done, verify your work against the success criteria as if you were a skeptical reviewer seeing it for the first time.'
The workflow I use to build these prompts starts in a separate Sonnet 5 chat. I enable a skill called Fable-Prompt-Generator created by Vanessa Chang (@ThinkwithV on TikTok), turn on the microphone, and brain-dump what I need. I ask Sonnet to keep asking me clarifying questions until it has everything, then generate the final prompt. That gives me a structured, complete Fable prompt without having to draft it from scratch.
When Fable has a well-built prompt, you can see the difference immediately in the planning phase. It breaks complex work into phases on its own, produces detailed tables of what it's going to do and in what order, and runs a self-audit at the end before reporting results. That self-check is where most of the quality improvement shows up compared to older prompting habits.
The takeaway for operators: Fable gets more useful as the task gets harder. If you are feeding it easy, narrow requests, you are wasting it. Give it your most complex, multi-step problems, front-load the context, and let it plan before it acts.
📰 AI News This Week
Anthropic Restarts Fable 5 After Export Control Lift
Fable 5 is back across Claude tiers after the Commerce Department lifted its export controls, 18 days after the model was pulled. Anthropic added a safety filter that blocks the flagged cybersecurity issue over 99% of the time, with paid plan users capped at half their weekly limits until July 7.
Joe's Read: If you're running AI workflows on paid Claude plans, expect reduced Fable 5 availability through early July and occasional false-positive flags on coding and debugging tasks until the filter stabilizes.
Meta Is Renting Out Its Spare Compute
Meta is reportedly building a cloud service to put unused data center capacity up for rent, with options ranging from raw compute to access to Meta-hosted models. The news pushed Meta stock up 9.3%, giving investors a second return path on the company's $182.9B infrastructure spend.
Joe's Read: For ops teams currently priced out of enterprise compute, Meta entering the rental market could create a more competitive pricing environment for AI infrastructure within the next 12 to 18 months.
AI Hits 16% on Professional Freelance Tasks
The Center for AI Safety and Scale Labs benchmarked AI agents against real freelance jobs graded by human professionals. Fable 5 matched or beat the human pro on 16.1% of projects, up from GPT 5.2's 2.5% rate when the benchmark launched in October 2025.
Joe's Read: For ops directors using freelancers for creative or design work, the more immediate impact is not replacement but throughput: your existing contractors can likely produce significantly more output using these tools, which changes how you scope and price that work.
🧰 Tool Worth Trying This Week
Claude Code
Claude Code is a desktop app that lets you build self-running AI workflows using a /goal command and scheduled routines. You connect it to tools like Gmail, set a goal in plain English, and it runs the task on its own without you babysitting it.
One practical use: pair it with Google's Design.md standard to generate on-brand website pages or prototypes without briefing a designer from scratch each time.
Caveat: This is not a polished point-and-click tool. If you're not comfortable installing apps and managing connectors, setup will take patience. It also runs on Claude specifically, so it won't work with other AI models.
🗺️ From The Field
The thing that surprised me most about Fable is that vagueness, used correctly, actually improves the output. That runs counter to everything most people have internalized about prompting.
But there is a distinction worth making: vagueness on the route is fine. Vagueness on the destination is not. Fable does not need you to tell it how to get there. It does need to know exactly what done looks like and what it is not allowed to break along the way.
This maps directly to how good ops handoffs work. You do not write a procedure that tells someone every micro-decision to make. You define the outcome, the constraints, and the escalation criteria, and then you get out of the way. Fable responds to the same logic. If you are writing prompts that read like a procedure manual, you are adding friction, not reducing it.
⚒️ Tools I Use
n8n — The automation tool I use to connect apps, trigger workflows, and stop doing things manually. If there's a repetitive process in your business, this is where you start fixing it.
VoiceInk — The brain-dump step in the Fable prompt workflow I described above starts with talking, not typing. VoiceInk handles the transcription locally on your Mac, so you can get your context and constraints out fast without anything hitting a cloud server.
Blotato — Handles the full content distribution side of your business: drop in a topic and it generates platform-specific posts, or feed it existing content and it repurposes it across formats. TikTok videos become tweets, podcasts become blog posts. Includes a scheduling calendar, visual creation tools for carousels and infographics, and publishes natively to 9 platforms with no per-post fees.
Beehiiv — What you're reading right now is published on Beehiiv. If you're thinking about starting a newsletter or moving off a clunky platform, this is the one I'd recommend. 20% off your first 3 months with my link.
Google Workspace — The Agentic OS that I built with Fable prompts is tied heavily to my Google Workspace. A Business Standard plan includes Gemini Pro built into every app and NotebookLM Plus, which makes it a practical hub for the kind of multi-model workflow this issue describes. 14-day trial and 10% off your first year.
Descript — Video and podcast editing that works like a text document. You edit the transcript and the media follows. Cuts filler words, cleans up audio, and handles captions automatically. 50% off your first two months on the Creator Plan.
💭 Final Thoughts
The broader lesson from this issue is that the tool is only part of the equation. How you hand off work to it matters just as much. Whether you are prompting an AI model or briefing a contractor, the quality of the output is almost always determined by the quality of the input: context, success criteria, and a clear statement of what is off limits. Get those three things right and you can afford to stop micromanaging the middle.
PS: If you want to know whether Fable 5 is worth the API cost for your specific workflows before July 7th, book a discovery call and we'll build the solution together in a live session.
Cheers,
Joe