[fragmede]

Honest answer is: whenever your markdown or json files get to be big enough that grep/jq takes long enough that you get bored waiting for it.

[embedding-shape]

> get to be big enough that grep/jq takes long enough

On a modern processor, that's about GBs of data typically, right?

[bitexploder]

Practically yes, but much earlier if agents are touching that data in my experience. Tens of GB even if you design well.