After refining the prompts and going through a couple of rounds of QC, the results of the auto review were overall impressive, especially if considered that the project contained documents in various languages besides English. The results were in line with DISCO's representation (about 80 to 90% exact responsiveness and issue tagging). For larger project this will end up saving substantial time and cost for the client. Support from the DISCO team was excellent. Review collected by and hosted on G2.com.
I am not sure this is suited for all projects. There is a hard cap (albeit very high) to what AI can review per document a soft cap for issue tags that can be created per project. If a project contains numerous very large documents (spreadsheets seemed to be the most common offenders) Auto Review might not be the best option. The time required to generate and refine the prompt ended up being longer than we had expected, but this is partially due to our inexperience with the Auto Review process. Review collected by and hosted on G2.com.