Smartly observing log of long running process

What feature would you like to see?

When I ask Codex CLI to execute long running process (e.g training models), after the log show the training process running consistently without error, Codex CLI just keep observing and waiting indefinitely without finish, so I have to manually Esc (not sure if it observing the log the whole time or not).

This is done much more intelligently in Claude Code. E.g:
"I see the training process started. I will wait a bit before checking again...
Run tool Sleep(30s)
I see the training process running successfully, the training log seem ok..."
And it finish

Are you interested in implementing this feature?

_No response_

Additional information

_No response_