Problem is, without modes, we lose capability that comes from the fact that modes exist, such as simple characters being compositeable commands;
How should a non-modal editor know whether that [d] I just pressed is me wanting the letter `d` or the delete command? It can't;
Either I have to somehow tell it "the next letter I press is to be interpreted as a command", or "interpret what I press as commands until I say otherwise". The former gets tiresome rather quickly (what if I have to do several commands?), the latter would again be a modal editor.
If you want stay in insert mode most of the time you could use Ctrl-o to switch to command mode for a single command and then be immediately returned to insert mode.
Eg
Ctrl-o de
To delete to the end of the current word.Again, I definitely wouldn't do it that way but vim makes pretty much everything possible.