"We used to check if Claude is doing the work right, e.g. by double-checking its output, catching when it stopped early etc. With Claude Fable 5, I instead check if Claude is doing the right work"
- Thariq (@trq212) Claude Code
"We used to check if Claude is doing the work right, e.g. by double-checking its output, catching when it stopped early etc. With Claude Fable 5, I instead check if Claude is doing the right work"
- Thariq (@trq212) Claude Code
Developers appreciate how Claude AI handles large code migrations, shifting focus from output checks to task selection so they can emphasize product direction and solve previously unsolvable challenges.
There are no real limits to doing things…

@sundeep true

@rohanpaul_ai @trq212 this is the shift. your job goes from QA to product direction. you stop babysitting the code and start steering what gets built
same experience here — I'm reviewing architecture decisions now, not catching off-by-one errors

@rohanpaul_ai @trq212 The shift from 'is it right?' to 'is it the right thing?' is the whole founder operating system in one sentence. When execution becomes cheap, the constraint moves to judgment. That's always been the expensive part.

@sundeep new skill we need to learn max ours limits

@rohanpaul_ai @trq212 Built an OS in Rust with Sonnet 3.5 recently. Great at generating, struggled fixing its own bugs. If Fable 5 self-corrects, that's the actual benchmark.

@rohanpaul_ai @trq212 That's a fascinating shift. The question is no longer "can it do the task?" but "did it choose the right task to do?"

@rohanpaul_ai @sundeep we've truly entered the era of ... just imagine it ...
imagine that you can solve the unsolvable and then try ...
just try and see what happens ...

@sundeep The only limiting factor.

that's a real shift in how you structure a review process. Less "did it execute correctly" and more "did it solve the right problem." the second question is harder to automate and easier to get wrong quietly, your QA process needs to move with it or you're just trusting out put you stopped reading
"We used to check if Claude is doing the work right, e.g. by double-checking its output, catching when it stopped early etc. With Claude Fable 5, I instead check if Claude is doing the right work"
- Thariq (@trq212) Claude Code