I'm trying to find the ortholog of protein A of species 1 in species 2 using BLAST. We also find additional homologs in species 2. Can we decide whether they are in-paralog or out-paralog using BLAST alone, or do we have to build the phylogenetic tree first?
Blast alone cannot tell you what kind of homology relationship two proteins share. Some researchers use reciprocal best hits as proxy for orthology (e.g., this paper), but this is more out of convenience than anything else.
If at all possible, I would try a tool like Orthofinder that is at least attempting to explicitly determine orthology and paralogy among your homologous proteins.