This is one of those videos which got me really excited. I mean I always knew you could lay down and arrange individual drum samples as audio instead of programming them in MIDI but I always figured it would be too time consuming and I never really thought hard enough about the creative possibilities. 
 
The kid in this video has convinced me otherwise. I now see a TON of creative possibilities and advantages to laying down drum hits as audio. Granted he's got his workflow down to a fine art and it would take a bit of practice to get this kind of flow going, but damn he's really made me realize just how much I'm limiting myself in beat programming by doing it all with MIDI and drum machines. Some very inspiring ideas here - even if you're not into EDM styles, and even if you're recording live drums and editing them later, I feel there's a lot you could pick up from this video in terms of drum arranging. Think I'll check out his other videos too because he's a great instructor. 
 

I like to leverage the best of both worlds, playing drums/perc in real time on a keyboard but recording it as MIDI rather than audio. If I want to do some fancy audio effect (e.g. reverse hit or reverse reverb) I'll freeze the drum synth, drag the hits of interest into a new track and mute the original.
 

An off-the-cuff comment in the newest Cubase promotion video that made me chuckle was "we try to make it to get MIDI to audio as quickly as possible." Reason I laughed was "Yeah, people can only hear audio, and MIDI FX suck... Audio FX are everywhere!" People only care about what they hear.
 
The advantage to MIDI (not mentioned in the video above), is note firing and tempo re-mapping. I.e., if I decide later on to fluctuate tempo around using the host's tempo map, clips all over the map are gonna cause extreme grief. Warping, glitching, etc. with tempo swings are just a can of worms that equates to pain. Also, if you decide later on that one sample needs to be "swapped out," you now need to reconstruct that track... straight tempo not so much an issue, but if you already mangled tempo, you are in for more pain.
 
That said, a "beat maker" that follows the host tempo, and has the tools internally to do what he showed is ideal. As I watched this video I had Geist in mind the entire time, and the only one I have not tried is his #6 - Reverb tails (8:10 marks in the video), although I think Geist can do this. I know you have Geist, because you are the reason I bought it .
 
The advantages to Geist (for me) are:

1. The patterns available allow me to move one to another, then edit the second quickly. Once a kit is constructed, the 24 patterns per engine allow "modified copy/paste" for variations that working in audio makes painful.
2. Tempo variations (avoided in the video above) - Geist syncs to host, so SONAR wanders, Geist wanders with it.
3. Swapping sounds out (also avoided) - Drag/drop new sound to a pad/layer, tweak it internally, done. The rest of the beat track is already updated. Linking audio clips would be the only "audio" alternative... good luck with that.
4. Internal audio engine - Geist covers many bases, but yes, you can record the output, or shoot sends inside SONAR.
5. Changes to warping/glitching - Everything done in this video was "once and done," no editing after the fact. Working with MIDI, you can make those adjustments "on tempo" far quicker, unless working with a VST/host that has that built in.
6. Creating a "machine gun stutter" - MIDI that has a sample gated on itself make this simple... he also did not do this in the video. Adjustments to such (i.e. loop construction window), no comment.

Quick Edit: Even the reverb aspect can be done by dragging a pad into SONAR, modifying it, and putting it back into Geist as a new sample... this does require the SONAR sample browser as an intermediary though, as the drag/drop can cause grief without a unique file name already assigned to that sample.

Good video.  It's super early here, but does he use the term "reverb barf" more than once?
 
One thing I noticed was how fast Ableton allows all of the sample manipulation to occur.  I'm sure he is a shortcut/keystroke genius, but damn.  Also, Ableton's drag-timestretch sounds 10x better than Sonar's.  However, Ableton's GUI makes me want to "reverb barf" though!
 
I agree with audio clips being easier and more straightforward to work with.  It's definitely "what you see is what you get," and it remains.  By "remains" I mean once you save, shut down and come back to it, it's still there.  No need to listen carefully for variations that may or may not occur that you like/dislike, no need to re-record, no need to load or constantly tweak channel by channel in a VST such as AD2, and it saves you from always changing something.  I don't want to think about how many hours/days/weeks I've wasted changing things, only to abandon a project!

It is an interesting video and well worth watching for some interesting creative tips.  A good watch!  I have always been a fan of converting to audio and while midi can also provide certain advantages over audio, at some point you have to convert to audio.  I find I start with midi, convert to audio and find myself editing audio after the fact a little further.  I use hardware synths so converting to audio early on once the parts are sorted out is also a good thing to do. It is nice to see the waveforms as well. It makes creating editing easier.
 
On the fact he is a power user and fast at using his program (Ableton) judging by the number of QWERTY key strokes he is making  (listening to the audio)  I can say that I do a lot of similar audio editing but on Studio One instead.  For me the workflow is a little faster and smoother again with that software.  Many audio functions are one key stroke (without modifiers as well)  but you can use the mouse a little more in editing on Studio One. The macros can perform a string of operations very quickly too.
 
Using the mouse and keyboard in conjunction with the Presonus Faderport also takes editing onto another level.

Recently I've been experimenting with doing a bit of Melodyne treatment to synth parts after I've bounced them. It's a great way to add a little interest here in there....for example you can take a fairly repetitive arpeggiated sequence and change/remove a few of the notes. And you can add and remove accents with the formant tool. I've have Melodyne 4 Studio and have yet to really dive into the overtone editor, again from what I've seen in videos a great way to shape otherwise bland audio to add interest. 

Probably goes without saying that this works better for some styles of music than others. For instance, MIDI gives me a lot more dynamic control, which is critical to some styles. But I think for glitch, dance, or other electronica, this guy's right on.
 
Great vid! Thanks for sharing.

Found it really inspiring, thanks Sharke. Having come from Trace and Drum and Bass ion the 80's I'm for ever fidning myself moving away from electronic synth music, although not completely.  I do want my synths to sound as organic as possible, rather than glitchy or mechanical. My recent purchase is an acoustic bass guitar. AD2 is my main drum source. I can see these tips working well in a acoustic set up. Used more sparingly than say a dubstep beat, but enough to add some other details to help the flow and fill some spaces, bring in new parts.Makes me think of when a drummer starts hitting parts of the kit that were not meant to be hit that way to add more textures to their performance
 
Being I use AD2, I'd have to sample it and put the sounds into a sampler such as Geist. I have enough tools now, so knowing there's and whole lot of options to play with in audio that can hark back to my electronica days is good to know.