I'm not sure what kind of machine is needed for doing boards at low volume. I think something that just puts down a few of the common caps could be handy, but on modern designs they're probably 0402 imperial or smaller so vision and nozzles need to be decent.
I wouldn't bother doing boards in-house unless I could put down most if not all the parts.
Future projects would likely be tailored around the limitations of the machine.
That's the best plan - consolidate components to reduce number of uniques, and then use the PnP for projects that suit it (design for that if possible) and outsource to a CM for designs that just cant work on the PnP.
Once you get to know your PnP really well and can get the time from "export from ECAD -> Setup and tune on PnP ->Running a board" down to sub 10-15min, then doing partial placements for prototype designs does become a viable option, but I'd imagine you're pretty fast at hand placement Dave
Though placing is quick.. getting parts out and not mixed up is the time sink.
If you are not going to feel guilty about the machine sitting idle most of the time, then no need to "find ways to use it"... if you are going to look at I and play the "regret game" about money spent, then finding ways to integrate it more into your day to day stuff is worth the effort of exploring.
I can get a new design from KiCAD into my N8 and placing parts on a panel (assuming no or super low feeder re-load) in less than 10 mins now, and much of that time is going up and down my stairs
Seon
Unexpected Maker