Acumen indexes content residing in the web directories to which it is pointed. Metadata can be in multiple types of XML, content can be of multiple layers of granularity and this is managed seamlessly. The metadata and content can be accessed and changed at any time without going through the software, which is really cool. Acumen uses the file names to infer relationships between the files; so if you want to organize your content by type of materials, then by collection within that, and then by item sequence within that -- reflect this in the file names, and Acumen manages it automatically. Example: images_jones1957_5.tif would be the 5th item in the Jones 1957 collection of images... and all the image collections would be grouped together for browsing. If a finding aid exists for this collection, it would be named images_jones1957.ead.xml. If this collection is a scrapbook, and page 5 has 2 images on it which you then want to present individually, the first one would be images_jones1957_5_1.tif and the second would be images_jones1957_5_2.tif. Metadata resides at the level to which it applies, and is named for the identifier (file name without extension) of that level. So if you have MODS metadata about page 5, it would be named images_jones1957_5.mods.xml. If you have Dublin Core metadata about the first image on page 5, it would be named images_jones1957_5_1.dc.xml. In our implementation, to locate content, you replace the underscores with slashes, and that's the path to the materials. It's not required -- but it's cool, that we can leverage the file system itself to organize our material. Plus this allows us to automate almost everything, which saves us a lot of money and time. And Acumen simply automatically indexes what we put in the directories, using the file names to make sense of what goes with what, and what is a subsidiary part of something else, and in what order. That's really neat.