A Million-Dollar App in the Fable 5 Window: Day 1 Log | Faceless AI - Dataspheres AI

A Million-Dollar App in the Fable 5 Window: A Developing Log Summary This is a developing log of an attempt to build an application with a credible path ...

A Million-Dollar App in the Fable 5 Window: A Developing Log Summary This is a developing log of an attempt to build an application with a credible path to one million dollars in value, using Claude Fable 5 as the primary engineer, inside the model's temporary availability window on the Claude Max plan. The log opened on July 3 against a July 7 cutoff. Since then the cutoff has moved and the build has stalled once and resumed; the running account of that is in Updates below, newest first. All work is managed under a spec-driven development system in which every task passes machine-checked validation gates before it counts as done, and every claim here carries a link to a verifiable artifact. The sections after Updates record the starting conditions, the method, the instrumentation, and what is demoable so far. Updates UPDATE — July 14, 2026. The window moved again: complimentary access is now extended to July 19, a week beyond the last deadline. The constraint that returned with it is the familiar one — I am out of Fable credits again, with the next refresh on Wednesday, July 15 at 8:59 in the morning. The runway is longer; the immediate work waits a day for credits. UPDATE — July 9, 2026. Back at it. With access now extended to July 12, the build resumes after a four-day stall. The tracked workstreams have in fact moved while the log was paused — eight of the nine are now demoable, detailed below — and the part the headline still owes, a specific application concept carrying the million-dollar claim, is the work the recovered window is for. UPDATE — July 8, 2026. The window moved twice in one day. Earlier word had exhausted Fable credits refreshing only on July 8, one day past the original July 7 cutoff, which made the exercise look moot: no credits to build with, and the refresh arriving after access was itself due to lapse. Then the access was extended to July 12 — which I heard from my wife the same day. The premise, which had looked finished, was viable again. UPDATE — July 5, 2026. Fable credits exhausted. The opening days produced real signs of progress across the boards, but the specifically-Fable premise stalled when the allotted credits ran out. Momentum, which the whole exercise depends on, stopped, and for three days the concept looked dead. Background: the availability window Fable 5 launched on June 9, 2026 alongside Claude Mythos 5, with subscriber access originally planned through June 23. On June 12 the U.S. government applied export controls to both models, and because Anthropic could not verify user nationality in real time, it suspended access for all users . The controls were lifted June 30, and the model returned on July 1 with two new constraints: a cap of 50% of weekly usage limits, and a hard end to complimentary access on July 7 — a date since extended to July 12 and then to July 19, as recorded in Updates ( CNBC , Search Engine Journal ). After the window closes the model is available only through usage credits, at $10 per million input tokens and $50 per million output tokens. Fable 5 availability: launched June 9, pulled June 12 under export controls, restored July 1, cutoff extended from July 7 to July 12, then to July 19. A frontier model that was export-controlled and un-controlled within a single month, and is now available on a deadline that has already moved once, presents an unusual measurement opportunity: a hard external cutoff against which one builder's output with the model can be recorded as it happens rather than reconstructed afterward. The moving deadline, and the credit exhaustion in Updates, are part of what is being measured. Working constraints for the build period. Objective and measurement The objective is an application with a credible path to one million dollars in asset value, built primarily by Fable 5, before the window closes. The figure is a valuation target for the asset, not a revenue projection for the window. Three measurement rules apply to this log: A claim is admissible only with a link to its artifact — a live URL, a board record, or an output file. Unverifiable claims are excluded. The log is published as the build develops, whether progress is fast or stalled. A stalled stretch — such as the one recorded in Updates — is logged with the same specificity as a productive one. Model usage is part of the record: what Fable built, where it failed, and what the work consumed against the 50% weekly cap, including the point at which the credits ran out. A closing update tallies the period's output against what it would have cost in usage credits. Method: spec-driven development All work runs under all-dai-sdd, the spec-driven development system described in Run AI Work Like a State Machine . The relevant properties for this log: work items form a machine-verified chain from a verbatim intake prompt through research, specification, and typed validation criteria to a cited artifact; no task advances without evidence; and a regr