There is a fundamental trade-off between coverage and gain. The most and least (respectively) you can get is a dipole or loop, which do not receive signals along the rotational axis. (A 1/4 wave whip is equivalent, it just has a ground plane replacing one element.)
There is no such thing as a true isotropic antenna, because EM waves are transverse and will always cancel out along some axis (which comes from what's called the Hairy Ball theorem).
If by "omnidirectional", you mean all directions within a plane, that's okay: place the dipole vertically (perpendicular to the plane of 'sight') and you will cover all compass directions.
You can even get some gain back, by using a vertically taller (e.g., colinear) antenna, which restricts reception to be closer to the plane. (The more precise you want the radiation pattern in a given direction -- in this case, closer to parallel in the plane -- the longer the antenna must be in the same direction. Again, trade-offs all over the place.)
ISM bands are not very wide (low bandwidth), so you will not need a large (wide bandwidth) antenna. An overly small antenna will cost you in terms of precision (needing to be tuned) and efficiency, so don't expect too much.
Tim