ChIP-exo does seem to be the "ChIP-seq killer." I've seen Dr. Pugh present it a few times, and the audience is pretty much always impressed.
One thing I'd do if I were of the "experimental bent" would be to add random degenerate barcodes in the library prep to control for potential PCR artifacts. I imagine that since the "peaks" in ChIP-exo seem to be quite a bit narrower than ChIP-seq, I reckon it'd harder to use some coverage-in-local-neighborhood to remove such biases (assuming you wanted to do so).
To give you an idea of what I'm thinking of -- it would be somehow similar to what is depicted here.