After chasing down a problem for longer that I care to share (conflict between splitter and regexp), I think it would be helpful if there was some visibility into the internal mechanics, at least at a high level.
For example, a way to show:
* Output of Filter processing
* Output of Splitter
* Output of Normalizations
* Output after Pattern Reductions
* Matching REGEXP
* Matching Model
You get the idea. This would greatly simplify tracking down SIML code that is disconnected from expectations.
It could output to the "Studio Console Debug" window or whatever.
And, since I'm dreaming, a 'trace' attribute on "significant" elements (ala RegExp and Models) with the intention that you are testing a certain model or Regexp, so when the Bot is traversing these marked elements additional trace is output.
In the long run I expect this would significantly reduce support questions.