I have a knowledge base that has multiple sources. These are different folders added to my Digital Ocean Space, and subsequently added to and indexed by the Knowledge base. Some of the files names are duplicated in some of the folders because they are older or newer versions of the same file. I want my agent to be aware of this and I want to question it about updates/changes to files with the same name but by default my Agent seems to have no awareness of the multiple data sources. It appears to see only one single large source/knowledge base.
What’s the best way to handle this on Digital Ocean’s GenAI Platoform?
This textbox defaults to using Markdown to format your answer.
You can type !ref in this text area to quickly search our full set of tutorials, documentation & marketplace offerings and insert the link!
These answers are provided by our Community. If you find them useful, show some love by clicking the heart. If you run into issues leave a comment, or add your own answer to help others.
The retrieval_info shows which knowledge bases and which sources were used if the original data was stored to digital ocean spaces with bucket and folder names that make the source clear.
In the JSON body of your request, include:
Your message(s)
The parameter
"include_retrieval_info": true
Hey @setec 👋
I believe that at the moment, DigitalOcean’s GenAI platform treats the Knowledge Base as one unified index. That means even if you upload files from different folders or sources, the agent doesn’t automatically understand that some files are older versions or that there are duplicates from different locations. It just sees all the documents as one big pool of data.
To work around this, you might be able to rename your files before uploading, adding version details directly to the filename, for example, including the date or version number. Another helpful trick is to include version or source metadata inside the document itself, like adding a header line that says “Version 2 – Uploaded from Folder B.” This gives the agent more context during retrieval. Also, when prompting the agent, be as specific as possible and mention differences between the sources if you’ve included them in the content or file names.
Also, I’d recommend reaching out to DigitalOcean Support to explain your use case and also submitting it as a feature request on the DigitalOcean Ideas Board. It’s a strong use case, and definitely something others would benefit from too.
- Bobby