You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A proposal has been made to introduce "RAG Bundles." They are intended to pre-select data sources, retrieval algorithm, prompt, as well as LLM or provider. However, there is a need for clarification:
What is the use case?
What would the user story look like?
Does such a bundle only make sense with external data (aka ERI servers)? The user already has control over local data sources.
If it only makes sense with external data, then choosing the data source would ultimately mean choosing an ERI server.
Currently, an ERI server connects to only one data source. Is that also the case with this proposal?
The text was updated successfully, but these errors were encountered:
After consultation on this proposal, it became clear that assistants were essentially meant. The idea is that an assistant builder can also create assistants that are connected to an ERI data source, etc. See #29 for the full description.
A proposal has been made to introduce "RAG Bundles." They are intended to pre-select data sources, retrieval algorithm, prompt, as well as LLM or provider. However, there is a need for clarification:
The text was updated successfully, but these errors were encountered: