Users highlight huge productivity gains from AI tools like Fable automating evals and site updates that once took nine days.
Based on 1 visible X reactions from 1 accounts; directional sample.
Ask a question below.
Published answers will appear here.
I also use codex to run the evals. Often there are quota issues, errors to clean, and so on. Letting an agent run evals and push site updates is a huge time savings. It would be difficult to keep up with the rate of model releases without the agent operational help. https://x.com/xlr8harder/status/2076244440734122214/photo/1
Models are better, but I'm also better at knowing how to use them, which includes how to anticipate and prevent their common failure modes. Fable used playwright to screenshot in-work pages, which reduced cycle time a lot. But it's visual capabilities are still limiting.
My personal effort has dropped a lot. My availability to work on this varies, but these times are at least indicative, and the results also improved each time. Apr 2025 launch: ~9 days Sep 2025 refactor: ~6 days Dec 2025 visual update: ~1 day Jul 2026 visual update: ~1 day
I also use codex to run the evals. Often there are quota issues, errors to clean, and so on. Letting an agent run evals and push site updates is a huge time savings. It would be difficult to keep up with the rate of model releases without the agent operational help. https://x.com/xlr8harder/status/2076244440734122214/photo/1
Models are better, but I'm also better at knowing how to use them, which includes how to anticipate and prevent their common failure modes. Fable used playwright to screenshot in-work pages, which reduced cycle time a lot. But it's visual capabilities are still limiting.
My personal effort has dropped a lot. My availability to work on this varies, but these times are at least indicative, and the results also improved each time. Apr 2025 launch: ~9 days Sep 2025 refactor: ~6 days Dec 2025 visual update: ~1 day Jul 2026 visual update: ~1 day
Users highlight huge productivity gains from AI tools like Fable automating evals and site updates that once took nine days.
Based on 1 visible X reactions from 1 accounts; directional sample.
Ask a question below.
Published answers will appear here.