m1sterlurk's 100% spot on here...this system has an awesome compliment of processors, but very little in the way of sources. While the Braids is a really great oscillator in of itself, having just one single audio generator isn't going to work well unless you're talking about a very minimal system build. Another problem that seems present is that some of these space-heavy processors are a bit redundant, notably the Morphagene and Clouds which are both granular sample/audio manipulator devices. If you were to let one of those go, you'd open up another either 18 or 20 hp (depending on what gets pulled out) and then that would give you more space for a bit more sources. Although, I'll second that vote for the Dixie II+, but put one of those 3-in mixers next to it to use as a handy waveform combiner to stretch out that multi-waveform oscillator's capabilities.
I'm also not a huge fan of Pittsburgh's first-gen stuff, so a different and more capable filter might also be a consideration. Hrmm...gimme a bit, gonna tinker with this...lessee...(exeunt stage left)
REMIX! C'est voila...
OK...I did some radical surgery on your original here. First up, I reworked the tile row's positioning to work together with the reconfiguration, putting the MIDI on the left above the VCOs. Buffers are next to this. Other buffer went away (redundant) along with the MakeNoise mult (too big, replaced with 2 hps for twice the mults in the same space). Modulation/S&H tile is now above the modulation sources, and the audio I/O is rightmost, again to follow the signal flow pattern with the processors now on lower-right.
VCOs are the Braids and two Dixie II+s which now have the mixers as waveform combiners as well as mixpoints for the Braids' output. This is a pretty critical rework, as it now gives you three signal sources, meaning you can use one for audio frequency modulation on the other two, and still have those two for a more complex audio source.
Audio flows down to a Morgasmatron, which is a very twisted take on the Korg MS-20's filter pair. The Pittsburgh VCF just didn't have the modulation inputs to do some really wild stuff that the Morgasmatron does easily, plus the M'tron is a dual-input VCF with a MIX output, which means you can also use this to mix audio as well, albeit with some strange stuff going on. Warps is next to tamper with the audio, either from the M'tron's MIX output or its individual filter outs, allowing you to further combine those down while processing them at the same time.
The Quad VCA is dead-center in the bottom, which allows you to easily use its VCAs for audio OR CV amplitude control. You can mix two audio signals together on VCAs 3 and 4, but still keep 1 and 2 free for DC-coupled linear work.
Morphagene is gone, as you'll note. As I noted above, it's redundant, and you don't want a lot of the same processing things in a small build like this. Kept the Clouds, which has the Erbeverb as its front end. An idea I had is this: take the mono feed from the Warps (1 + 2), feed that to the Erbeverb's mono in, then use it to generate a stereo field to feed the Clouds with, and the Clouds' output goes direct to the I/O above. So, technically, a big chunk of the middle of the audio chain is where the mixing gets done, in increments, until you hit that last processor pair and tamper with the mixed audio plus create the stereo field signal to go straight to the outs. Also, bringing the inputs in is a bit easier now, since they can come in either in mono to VCA 3 while VCA 4 handles the generated signal path...but that's just one method, and this creates several, now.
Above the processing is modulation: a dual ADSR to better deal with final VCF/VCA envelopes, the Maths for loads of definable modulation, and a Batumi for four fixed-waveform LFOs. Plenty of modulation signals now to feed the various modulation-hungry changes!
Anyway, this is how I'd approach the situation. Yes, it means swapping some things out and doing some radical surgery, but this is a much more capable build, starting with the bones of yours...and it had pretty good bones, just needed some reworking. Another suggestion (which m1sterlurk's on below) is to study experienced synthesists' builds, and glean information from seeing things that those people tend to do in common...because they work. It's also best to do and do and redo and redo and redo again and again here on MG before pulling the fiscal trigger for a system, because you can both study up AND learn how to make, then avoid making, mistakes.