Now we're building something. Strong enough to carry a sprint.
Optimistic UI, streamed output, and a loading state that reports progress instead of hiding it.
So we don't start with one. Day 1 of a sprint is this: write the post a real user would publish if we got this exactly right. Then we work backwards from the sentence until we know what has to be true for someone to actually write it.
Click anything on the card to rewrite it. The chips underneath are what that wording just committed us to building.
Now we're building something. Strong enough to carry a sprint.
Optimistic UI, streamed output, and a loading state that reports progress instead of hiding it.
Everything Day 2 builds has to earn one of these sentences.
This is why the sprint isn't a template. Two clients write two different sentences, and the build diverges from that first hour — one spends its budget on latency, the other on the one screen that gets screenshotted. The method is fixed. What it produces never is.