[felisml]

If you tell me what your thing does, I can probably get a working prototype together within six months, even absent any more technical detail than "applying deep learning to audio processing." That's the nature of the thing. Once you show people that something can be done, they can probably figure out a way to do it themselves. If they really wanted to, they could probably even toss a bunch of patents in your direction and work out the specific details later.

So while I would avoid spelling everything out & focus on practical / business details...you're mostly worrying about the wrong stuff.

PG's essay about corpdev might be worth your time. http://paulgraham.com/corpdev.html