Startup Ideas Bank
Translating AI's Black Magic into a Playbook, but the Market is Still Figuring Out the Spellbook.
AI roast score: 72/100 (B)
The idea
Sahir619/fable-method — The Fable Workflow: how Claude Fable 5 worked, distilled into skills any model can run, with the eval that keeps it honest. Think / act / prove.
The Fable Method (The Fable Workflow)
How Claude Fable 5 worked, written down before it was gone. With the eval that keeps it honest.
In its final days before getting removed from the Subscription, Claude Fable 5 distilled its own way of approaching problems into a set of skills any model can run: classify the ask before touching anything, define done with a named verification, gather evidence in parallel from primary sources, commit to one recommendation, change the smallest correct thing, verify by observation, report the outcome first with honest caveats. Then it tested that distillation against itself, adversarially, across fifteen eval rounds and more than 260 agent runs, and kept the failures in the log.
Most agent instruction files tell the model what to value ("be careful, verify your work"). This one tells it what to do, in what order, with thresholds , so a mid-tier model can follow it literally. Four skills, one philosophy: think (fable-method), act (fable-loop), prove (fable-judge), grow (fable-domain, which generates new domain adapters the way the author model was observed making one). Every rule exists because a test failed without it or a trace demanded it; every claim below links to the committed transcript that backs it.
Results at a glance
Fifteen eval rounds, more than 260 agent runs, blind LLM judges that verify by diffing and executing, never by reading reports. Read the evidence as stories: eval/cases/ has one case study per scenario (the exact problem, what each agent actually did, who passed); start with the surprise trap . Full log: eval/RESULTS.md · raw judge outputs: eval/results/
What was measured
Without
With the method
Evidence
Haiku surfacing a spec-vs-test conflict instead of silently "fixing" correct code
0 of 4 runs
4 of 4
round 3
Sonnet on the same trap
flags it, then sides with the wrong test
ideal action, both runs (8/8)
round 3
Sonnet vs a bare frontier model across code, data, and research problems
n/a
ties or out-ranks it on 3 of 4
round 4 , round 5
Haiku catching planted frauds in a lying "work complete" report (fable-judge)
4 and 3 of 5
5 of 5, both runs
round 8
Haiku finding the brand-rules and product-facts files before judging marketing copy
1 of 2 runs (one run praised a fraudulent price)
2 of 2, 6/6 frauds both
round 9b
Bare Fable 5 itself resisting an unauthorized staging deploy that the fixture's own README prescribed
1 of 2 runs deployed unbidden
the authorization gate exists because of this run
round 11
Sonnet generating a research-gro
The roast
The ambition to distill the proprietary 'magic' of a sophisticated AI model like Claude Fable 5 into a replicable workflow is commendable. You've clearly invested significant effort in documenting and testing this 'Fable Method,' and the provided evidence of its efficacy in specific scenarios is compelling. However, the core problem remains: how do you commercialize a methodology that is inherently tied to the underlying AI model's capabilities and availability? The market for 'AI workflows' is nascent and dominated by tools, not just processes. Your 'q15 biggest unknown: will_pay' is the elephant in the room, and 'q12 stage: validating' suggests you're grappling with this fundamental question.
While you've identified a problem of AI agents failing to follow instructions reliably, the solution of a 'method' might be too abstract for broad adoption. 'q5 audience: developers, professionals' and 'q4 buyer: enterprise, smb' suggest a wide target, but the value proposition needs to be crystal clear beyond 'how Claude Fable 5 worked.' The transition from a documented internal process to a scalable, subscription-based product ('q7 revenue: subscription') is a significant leap, especially when the underlying AI technology is not proprietary to you. You're essentially selling a better instruction manual for AI that others could potentially replicate or that might become obsolete as AI models evolve.
Red flags
- The core 'method' is tied to a specific AI model's observed behavior, making its long-term defensibility as a product qu
- The primary value proposition is a documented process rather than proprietary technology or a unique dataset, making 'q1
- The 'q15 biggest unknown: will_pay' is a critical hurdle that isn't adequately addressed by the current documentation-ce
Verdict
Focus on building a tangible product around this methodology, rather than just the methodology itself, to address the market's need for actionable AI tools.
Roast your own startup idea →