Yes, the UMIs are random 6 nt sequences that will be added to all samples. Therefore, the UMI sequence cannot be used to uniquely identify the different samples – only to tag distinct random priming events during library generation.
Learn more about UMIs, their function, and usage in our Lexicon Chapter #8.
The i7 (and i5) indices are still required if multiple libraries need to be uniquely barcoded and pooled for sequencing in a single lane or run. The i7 and i5 index sequences are added during the endpoint PCR and read-out during sequencing with index reads 1 and 2, respectively. The index reads are required for sample demultiplexing after sequencing.
...