This is more of a comment, but due to the lengthiness I am posting it as an answer, which it isn't.
If by protein family you mean more than the 3-5 similar protein in bony fish and tetrapods resulting from the x4 genome duplication after the split with lampreys, the solution by @vkkodali is perfect for you —and I think that is the case. Doing a Blast of the gene with
Petromyzon marinus as an outgroup also works well especially if you want to infer some relationship between them. The huge ABC transporter family is one of the exceptions, so this trick won't help as it duplicate many many times across the tree of life.
If you meant domain fold family, i.e. in the structural sense, like TIM barrels, Rossmann folds, β-propellers (WD40 repeat), SH3 and so forth, you are after PFam families, for which Uniprot is a much better database than NCBI, albeit trickier to parse programmatically: uniprot: P33897 gives Pfam: PF00005.