Only 11 of 340 AI outputs shipped, and review is now the bottleneck
TickenChikka2 · reddit · 2026-07-25
The author tracked a month of AI-generated drafts, images, scripts, and snippets and found that only 11 out of roughly 340 outputs were actually shipped, sent, or shown to a client. The post argues that this is not necessarily waste: when generation becomes cheap, the bottleneck shifts to human review and selection.
Key points:
- Before AI, the author made only a couple of options because generating more was expensive.
- Now the hard part is not prompting, but reviewing enough outputs without burning out.
- The real constraint is “review capacity,” not model quality.
- The next helpful tool may be something that helps discard faster, or reduces the number of things a human needs to inspect.
More from Apps
- YC F26 startup AnyAlpha launches agentic trading platform that turns ideas into 24/7 trading agents — Scobleizer · 2026-10-27
- Anthropic Expands Claude for Nonprofits as 25,000+ Organizations Adopt Claude — claudeai · 2026-10-27
- Claude Projects now lets you drag threads into the message box or @ mention them for precise context — dfeinition · 2026-10-27
- Higgsfield launches Playable: turn a single prompt into a playable mobile game — CurieuxExplorer · 2026-10-27
- Open launches Open for Sales: one prompt to build voice, email and WhatsApp AI campaigns — ycombinator · 2026-10-27
- Rork streamlines app screenshot localization across 10+ languages — rudrank · 2026-10-27