So long as you don't mind that your ClientIDs will have gaps and not
necessarily always be postive, etc., then there is no problem using
autonumber as a ClientID.
What usually occurs, though, is someone begins to use autonumber as a
sequential numbering device, and then decides that all numbers must be
consecutive -- no gaps allowed. And then the person decides he/she wants to
change a number for an entity, and wants to know how to reassign all the
numbers (which should not be done in any situation involving the primary
key, regardless of whether you use an autonumber or not as the identifier).
It's possible to use the autonumber as the "hidden" primary key and then to
generate (even automatically) a visible ClientID that can be enforced to be
consecutive, with no gaps, and even could be "changed" if desired -- all
without affecting the very important primary key value.
So our initial recommendation is always to not make the autonumber the
visible value IF that value is to have "value" (sorry, couldn't resist the
word play!) to the user. But it can be used for this purpose so long as the
database developer and the user understand its inherent capabilities.