What are some best practices for managing historical documents?

Sarah Busse

We are trying to figure out the best way to add historical documents into our library so they are beneficial to the AI search, but also make it clear that these items are not approved to include with any responses. For example, we keep a running log of all major events in the company, which can be useful when answering questions about awards or about location history, but the documents themselves should not be included in any formal response. I've thought about star rating, but if I give the historical ones a low star rating, I'm concerned it will not include them in the search. I've also thought about creating a separate tag, but again, I'm concerned it won't be clear that these are not for outside use. Do you have any suggestions on how to approach this?

As a secondary question, we also have instances where documents are updated, but the previous version could also be helpful. Can AI search previous document versions, or would these need to be loaded as separate documents?

Zach Leitch

Great question, Sarah!

I think my initial recommendation would be a private collection, with very limited access, for all of the historical content.

That way, you and a select group will always have the ability to include that collection in AI related processes, content generation, etc. when you want to and to easily filter it out/exclude it when you don't.

BUT

The average user won't have access to that content and therefore won't be able to mistakenly use any of it in responses, etc.

Happy to brainstorm more but also very curious to hear how other teams have done this!

Please chime in if you have any insight, thoughts, etc.

Regarding your secondary question about AI searching previous versions of documents: I'm pretty sure AI would not consider previous versions, as you generally wouldn't want it referencing out of date information, but I'll confirm with product and let you know.