Sabbagh and colleagues frame “near-perfect genome sequencing” as the convergence of long-read sequencing, diploid genome assembly, pangenome references and AI-assisted interpretation. The goal is to replace today’s fragmented diagnostic cascade and close short-read blind spots, including repeats, segmental duplications, complex structural variants, methylation and phasing. They propose that genomic completeness itself should inform variant classification, including VUS assessment. The one-test model could reshape postnatal, prenatal and cancer genetics, but cost, computation, equity and ethics remain key barriers. (https://www.nature.com/articles/s41588-026-02645-4)