Earlier quoted context omitted.
This treatment does not use CRISPR. It involves the hollowed out and repurposed lentivirus (similar in kind to HIV)[1]. The virus keeps it's own viral 'insert-into-DNA' machinery, but is stripped of its replication machinery as well as the code for its physical shell. The insert-into-DNA machinery is further hijacked so it can insert nothing but the DNA that encodes for the new mutated hemoglobin (HBB [T87Q][2]) that…
CRISPR is typically delivered via a lentivirus.
CRISPR (more specifically, Cas9) is another protein machine that locates particular DNA sequences [1]. It can help take over from the lentivirus with respect to the 'insert into dna' capability. But you're right, we still have no better way to get it into the cell than to use a (lenti)-virus's own 'insert-into-cell' machinery. So a lentivirus might be used along with CRIPSR in order to just get CRISPR machinery into the cell.
This is further muddled by the fact that sometimes you actually want the dna that encodes for your CRISPR system to itself be inserted into the genome, and in that case you might keep both the lentivirus's 'insert-into-dna' machinery AND the CRISPR's own 'insert-into-dna' capability.
At the end of the day, lentivirus's capability to insert into DNA is not predictable, and therefore a bit dangerous (if it inserts its payload in the middle of an oncogene in a predisposed cell you could get cancer). CRISPR is a device which promises to bring specificity to the command, making it 'insert-into-dna-AT'. And in that way it could replace the job of lentivirus in the above gene editing technique.