Hey Greyg Sinigaglia de Malibran, facing similar challenge in our company, albeit we use Google Drive + Confluence.
I only joined the company 2 months ago but my role is to improve and further implement DUST (among other things). It was clear from day one that both data sources include super outdated documents and the agents do not know how to prioritize them - which makes sense.
I haven't gotten around properly addressing this other than prompting agents to prefer more recent documents & searching the database with date cut-offs.
What you suggest sounds fairly nice although it won't be that easy to implement. From my point of view, I started creating agents used for very specific narrow use-cases, for these I select the sources carefully and so the issue with outdated information does not occur. It only occurs when employees use the base atDust agent. So at least for now, and at least for how we envision going forward with Dust, it's about creating narrow helper agents for specific tasks where the issue you described is solved at the creation point, where we carefully selected the truly relevant and updated sources where owners know that this will be the source going forward and they should keep it up to date. It's to be seen if this will work in the long run...