I would find the addition of user-definable “fields” to documents, including the capability to do simple math computations for items such as receipts, useful. Christian has hinted at plans for a future generation of DEVONthink.
There are several ways a bit of number crunching can be done within DEVONthink.
DEVONthink has a special document kind, Sheets, that can emulate a simple spreadsheet, do some simple calculations and swap data with applications such as Excel or Numbers.
The Calc Service allows some calculations to be performed within text documents.
The workaround I use when tax time rolls around is to include within my financial database an Excel sheet (sometimes more than one), that’s available for helping sum up receipts, income, etc. that’s backed up by the documentation (copies of receipts, 1099 forms, interest income, distributions, proceeds of sales of investments, etc.) needed to satisfy IRS.
By comparison to the workflow if I were using Neat or Paperless, I don’t enter numerical and other data into fields at the time I scan paper copy such as receipts or 1099 forms. Using my ScanSnap and DEVONthink Pro Office, I just dump paper into the ScanSnap’s feeder and scan copy in without any interruption of the feed/scan/OCR process (I don’t even want to be interrupted by entering metadata such as Title or Subject, and turn off that option in DEVONthink > Preferences > OCR). So my workflow for capturing paper into a DEVONthink database goes much more quickly than if I were using Neat or Paperless.
My workflows involve steps after scanning, rather than upfront. The first step is to Name and categorize each document after capture (including filing and perhaps tagging). The second step is to add necessary information for processing the numerical data, e.g., when that becomes necessary at tax preparation time. I’ve got all my receipts organized, and simply copy/paste or type data extracted from each into an Excel (or Numbers) sheet area. In a different sheet or area, enter data from 1099 forms, etc. So instead of a priori entry of metadata about a document at the time or scanning, I do that a posteriori, which is quicker and much more flexible – and requires less typing. As I’ve got the data in a spreadsheet now, it can be flowed into tax calculations, especially those involving special cases, more readily than data from Neat or Paperless.
But such financial documents represent a tiny fraction of the documents in my databases, and the vast majority of my documents are not scanned in. I would rebel against DEVONthink if it required me to do a priori entry of fields at the time of capture of each document – I sometimes enter new content in batches of hundreds or thousands – as (with hundreds of groups) categorization by field entry would be a nightmare, and would ignore the features that I actually use (including Classify, See Also, searches, smart groups) to save me time in managing and using my collections. Needless to say, Neat and Paperless are very limited in overall database features and powers by comparison to DEVONthink, and depend heavily on data entered by the user as each document is captured.
So yes, I would find the ability to create metadata fields useful. No, I don’t want to be required to do that at the time of capture of each document, and wouldn’t apply such fields to each and every document in my databases. In that sense, the current lack of user-definable fields is not an “obvious” deficiency in DEVONthink.