We are currently working on a single-track sequencer but are playing with ways to make a multi-track machine that retains most of the fun of this WYSIWYG machine. The realisation is that once you go multi-track and more importantly: if you have a directed musical approach (ie. you know what you want to program in), the DAW is extremely hard to beat. It's always faster in editing 'whatever.' A HW UI to enter 'any kind' of data tends, really, towards the fast, generic computer mouse/KB/shortcuts once you need that hardware to enter in all kinds of data and not just one parameter per one hardware element. We used 2 OTs for sequencing and even that was just wasting time once we reached that threshold of 'I need to do these specific things to the sequence.' Even more importantly, this HW still had a ton of constraints - which you really don't want when you're getting closer to the completion of a song because of the directed nature of the work during this side of the process.
I would say use hardware for the creative curation fun end of things. To do sequences you can't think up on your own or create your own sequencing paradigm via patching in a modular environment. We've been down this road and done lots of testing and prototyping trying to do DAW sequencing in HW and it ends up being a compromise not worth putting in HW because you either make an enormously huge machine (to which point the UX begins to fail for ergonomic reasons) or you make a generic 'layered' input method. We're still working on a way to do a multitrack seq but there are some severe restrictions you have to place on the machine - without those restrictions, we'd rather just 'prototype' one up in Max to do what we like or just go right to the PRV in a DAW. We are designing in a way that says the DAW is still a part of the song making process. And this unburdens the design significantly. The question is really how far into the workflow we can go with a HW sequencer before it falls apart trying to do too much or do things in a worse way than can be done in a DAW. We also want to make sure this 'jump' is somewhat up to the user by making sure connectivity to a computer is quick n easy.
There is a workflow crossfade most musicians need to go through:
Creation/inspiration/fun <x>'Work' doing the nitty gritty to complete a full song.
The former is where you want creative HW. The latter is where, for us, the DAW kills it. There is a similar crossfade layer (well many, really) from left to right in terms of the balance between:
Open-ended 'suggested' ideas <x> "I want this exact thing."
The nice thing is that with fun hardware, you listen and curate more purely because you are operating in a liter headspace - because it's fun to use creative HW moreso than a computer/mouse/screen. We spend a lot of time on a computer anyways - esp with all of the non-musical CAD, simulation, and other technical design work. So breaks from it are quite appreciated and inherently fun! We keep this mental space sacred and our machine strives to have this reverence too!
If you have to stay entirely in hardware (most folks want the potential reliability/simplicity of a non-computer-driven approach), have a look at more comprehensive solutions too! Things like the Cirklon or even the OT (because what they do to audio is great on the creative end!). Another trick we used to do is to play into these machines with a keyboard or even fire off MIDI from a DAW and record it in.
Maybe you can think of your own, personal workflow crossfades <x> and see how you would like to work throughout the process. For us, our own preferences and understanding them really helps not only design of new machines but also what works for what when making a song.
Have fun!