Video summary

I Asked Fable 5.1 and GPT-6 Astra to Get Me Out of Copy Paste Hell. The Results Surprised Me.

Main summary

Key takeaways

Product Review

Product Reviewed

An AI-built native Mac clipboard manager (two variants) designed to fix “copy/paste hell” by remembering text, images, and links—so you can restore copied items via a keyboard shortcut rather than managing recency manually.

What Each Model/App Built (Core Features & UX)

“Shelf” (Astra / GPT-6 Astra)

  • UI layout: A wide bottom shelf with big cards and a larger preview pane
  • Precision: Slightly less “pixel-perfect” than a list, but offers more space for inspection
  • Hotkey UX: Opens via Control + Shift + Spacebar (noted as easy to remember; big spacebar is intuitive)
  • Feedback/confirmation: Shows a “copied” success message
  • Keyboard behavior fixes:
    • Adjusts hotkeys that interfered with other macOS actions
    • Waits until you release the key before returning focus to the prior app (prevents accidental triggering underneath)
  • Iterative improvements: After follow-up prompts, versions 1.1 and 1.2 added:
    • drag/drop
    • improved shortcuts
    • confirmation messaging
    • more reliable keyboard focus
  • Installability: Produces a signed macOS DMG installer (authorized for legitimate installation)

“Ledge” (Fable 5.1 / Claude Fable 5.1)

  • UI layout: A narrow right-side list (“ledge”) that slides in from the right
  • Precision: “Code looks like code”; shows details like color swatches clearly
  • Hotkey UX: Uses Command + Shift + V (works fine, but the reviewer found Shelf easier/more intuitive)
  • Usability issues called out: Less usable due to “tiny little details,” especially:
    • hotkey/keyboard handling problems
    • less refined micro-interactions compared with Shelf

Pros (As Stated)

Shelf (Astra)

  • More usable immediately: the reviewer “actually used” it
  • Faster iteration: reached versions 1.0 → 1.2 quickly (Fable required longer early on)
  • Lower token usage: saved about an hour, allowing more testing and iteration
  • Better confirmation: clearer “copied” messaging reduces mental load
  • Fewer keyboard conflicts: shortcut interference fixed; focus release timing fixed
  • More customizable/adjustable: drag-and-drop and repositioning supported after updates
  • Better visual experience: cards + preview preferred by the reviewer
  • Real-world validation: an extremely busy user installed immediately and said she’d love to use it

Ledge (Fable)

  • Cleaner, more precise presentation for some content (text/code feel, image/color swatches)
  • Still liked the design
  • Follow-up questions during build: helped clarify requirements like which screen edge to use

Cons (As Stated)

Shelf (Astra)

  • Initial design had alignment/interaction issues:
    • cards not lining up perfectly
    • needed positioning/drag support
    • hotkey conflicts
    • lack of earlier feedback (later fixed in 1.1/1.2)
  • Wider layout and less “precision-first” than Ledge; may be less ideal for users who only need quick single-item grabs

Ledge (Fable)

  • Hotkey and usability micro-details ultimately made it less usable than Shelf for the reviewer
  • The reviewer emphasizes that these small frictions matter for daily use

Comparisons and Evaluation Criteria Used

  • Both apps:
    • came from the same brief
    • got the same time
    • produced functional tools
  • The key comparison wasn’t only “which model is smarter,” but:
    • how they behave in the second round (iteration after the first build)
    • how quickly you can go from prompt → usable app → refined build
    • how friction differs in practice (keyboard behavior, focus timing, confirmation messaging)

Ratings / Numerical Scores

  • No explicit numerical ratings/score were given.
  • Quantified statements mentioned:
    • Astra built improvements in time it took Fable to build ~1.0 (reviewer’s timing comparison)
    • Astra saved about an hour (token usage)
    • Version milestones:
      • Astra reached 1.0, 1.1, 1.2 quickly
      • Fable lagged in reaching comparable detail early
    • Astra-installed build: “passes over 65 different checks” covering clipboard restoration, saved history, keyboard behavior, etc.

Overall Recommendation / Verdict

Recommendation: Choose “Shelf” (Astra) if you want the most practical, low-friction clipboard manager that feels polished for real daily use—especially due to faster iteration, better keyboard/focus behavior, and clear success feedback.

“Ledge” (Fable) is still a strong UI concept, especially for precise content viewing, but it lost due to micro-UX details that matter for speed and confidence.

Unique Points Mentioned (Consolidated)

  • Two distinct UI paradigms from the same brief:
    • Shelf: bottom wide cards + preview
    • Ledge: right narrow sliding list
  • Importance of iteration (“second and third round of prompting”), not just the initial build
  • Astra:
    • faster iterations
    • fewer tokens
    • more rounds tested
    • better real-world usability
  • Ledge:
    • liked the design
    • usability hurt by hotkey/keyboard details
  • Hotkey differences:
    • Astra: Control + Shift + Spacebar
    • Fable: Command + Shift + V
  • Shelf improvements after feedback:
    • alignment, drag/drop, repositioning
    • conflict-resolving hotkeys
    • confirmation message
    • focus timing fixes
  • Astra output included a signed DMG installer for easier distribution
  • Real-world tester installed immediately and wanted to use it
  • Broader guidance from the reviewer:
    • use both models for different tasks
    • Fable for deep thinking
    • Astra for structured execution/design/building
    • Astra for “computer use” workflows
    • multi-agent systems not required for this build

Speaker Views

  • Single primary speaker (Nate, implied by context):
    • Provides the main comparison between Shelf vs Ledge
    • Emphasizes speed/iteration, usability friction, keyboard confirmation, focus handling, and real-user validation
    • Adds broader model-selection guidance (writing, spreadsheets, computer-use tasks, multi-agent systems)

Original video