The plan is to start with a vending machine just so we can implement new core features and do real world testing, then we move to gas pumps, and merchants(food, rent, clothing...).  The goal is to improve the experience vendors/merchants will have adopting dash.

We are adding features on this first product - multiple product selection, inventory control, serial connection software, better single board computer(Odroid-XU4).  We are also customizing the daemon so it can work faster with slow CPUs, along with finding ways to limit data usage.  More advanced wallets with multiple product selection ability.  It would be too much of a leap to go to a retail store until we can test IX and the software and wallets in a limited risk environment.

I think we can take advantage of the coin changer as a way to do a Dash purchase ATM, but I don't have all the details and am not promising anything.  Only that if it is legal we will figure out how to make it work.  The principle of an ATM is really a stop gap solution.  Bitcoin and it's slow confirmation time can only work if it has ATMs to do point of sale.  Bitcoin won't have point of sale adoption because the merchant won't take the risk of double spend 0 confirmation transactions.  Dash has InstantX so it doesn't need ATMs to get to point of sale.  Do we need a Dash to Fiat exchange, yes.  How much easier do you think it would be to get an investor if we had some concrete examples of Dash products in use?

There are a few ways to look at why any merchant would switch to Dash.
#1  Use a 3rd party like bitpay to instantly convert Dash to fiat at the merchant.  Sounds easy right?  There is little advantage to this scenario.  The fees(now or future) will still be high.  The user experience is now pushed on to Bitpay and how fast their system is.
#2  Use a debit card preloaded with Dash.  There is no advantage to this scenario.  It would be simpler to just use the debit card with Fiat.  The fees are hidden in converting fiat to Dash or with non use fees.
#3  Accept Dash directly.  This gives the most advantages.  ~0 fees.  Fast transactions.  No charge backs.  Unless we actually make this easy, #1 or #2 will get rolled out and the advantages to using Dash start to be limited.  Sure this is a big leap, but the benefits are huge.  Plus it opens the options for new features.

Now what if we add these features they don't have today?
#3a Dash Fast Gas.  You would pay via wifi through your phone.  Pay with your phone, pump and your change is returned.  The advantage here is we bypass all of the credit card payment stuff.  The merchant doesn't have to deal with charge backs, drive-offs, or fees.  The user advantage is that they can use their phone to pay without getting out of their vehicle(like if is 10 below outside).

#3b Dash Fast Lane.  No registers, no space for lanes, all product scanning and paying done with your phone.  Now any merchant can accept dash without even using a complicated POS system.  This is what we want.  This is the type of product that users will choose for faster checkouts and merchants will want because they don't need to waste space with register lanes or pay credit card fees.

There are many more products(including some 'interesting' ones) than just these, but this should give you and idea of what is possible and where we want to go.  We are creating the technology, not focusing on specific products.

The other item that I am maybe not making clear, the goal of Dash shouldn't just be to make a computer to computer transfer system.  We need it to be used outside of an office.  We were able to implement mobile IX sending because we made something that needed it.  I see similar advances here.  If we focus on merchant usage, the Dash features and functionality will come up to speed much faster.