Full content is useful for GPT doc search use cases but comes with issue that some might be too big to be returned. However for some customers (including Visualsoft), their data might be of a suitable size to allow processing of full content.
Would it be possible to have a switch on the doc search action to 'include full content' along with an appropriate warning to use at your own risk and may cause errors etc?
Should default to off so we don't bring it back by default.
We have tested the concept against Visualsoft's data with a custom Azure index that should be very similar to the Tenjin index but which allows us to return full content for GPT to answer questions from. They liked the process so after a POC (using the custom index) may well want to run with it but we need a Tenjin feature to go into production.
