Low drift is only one of many parameters than can be optimized. If you need low drift (usually expressed in terms of microvolts per degree C temperature change referred to the input) then you should use such an amplifier. If your requirement are for extreme accuracy at normal voltages or to deal relatively accurately with tiny relatively slow-moving voltages (such as thermocouples), then you may need a low drift amplifier. There may also be clever ways to avoid it.
However there are always trade-offs. An amplifier optimized for very low drift may have weirdness such as pulsed current coming out of the inputs. It may be expensive and/or unavailable. It may require voltages you don't have available or it may not be able to handle high enough voltages. An amplifier optimized for low audio distortion may not be very good with small DC voltages and may have relatively huge input bias currents (say in relation to your 100uA).
You really have to understand your application and what all the requirements are to do a proper job of design.
The LMV324 has maximum offset voltage of 7mV and typical (no guarantees) TCVos of +/-5uV/°C. So if you are dealing with a 2V voltage, the offset represents a potential error of +/-0.35% and the drift for 50°C change is only +/-0.013% typically. For a lot of general purpose applications, that's quite a useful component. Unfortunately it can only work from 5.5V maximum and is pretty slow. It's cheap, but not the cheapest. It has reasonable input bias currents, but much higher than a CMOS op-amp etc. etc. It represents National Semiconductor's attempt to chew into the market share held by their LM324 (and second-sourced by many competitors), with the main advantages gained being rail-to-rail output swing and lower supply current, while retaining the input range down to the negative rail, and improving on the upper end to 1V below the positive rail.
