I think that's one of the most important things about MVP - don't bother automating processes until they are actually taking up your time.
As a consultant I see a lot of people put way too much effort into developing bespoke software to automate or "improve" workflows that don't even really exist in their business yet, only to find out that the real bottlenecks exist in places they never even knew (and that they've wasted a bunch of money building something they'll never really use).
It's like a "real life" analogue of the adage "premature optimisation is the devil".
Similarly, we started out in my cofounder's grandmother's house and ran our original site on a $20 shared host running a horribly unstable cgi-bin with perl. My cofounder took calls and fulfilled orders himself.
