Document retrieval/Context?

Support
1

I’ve read the support docs and am having trouble understanding how to carry over content (if that’s even possible), so Narrova knows the content of previous scenes, so it can match detail (not just tone/style) in a new conversation. I’m working in Story / Storyforming drafting scenes, and have uploaded to Conversation Content a docx document containing the relevant scenes.

Narrova tells me:

If you tell me what interface you’re using (Narrova web app vs API chat vs something else), I can also tell you whether there’s a mode where document search is available and how to use it.

I’m using Narrova web app.

  • Goal: get me to actually search/quote your uploaded docs in Narrova

  • Reality check: not every Narrova chat has Document Retrieval turned on

  • Next step: verify whether your current conversation is “doc-aware”

  • Fallback: if not, paste excerpts and I’ll still tone-match/continue cleanly

In the Narrova web app, you can only get true “reference the upload directly” behavior if your chat is running with document retrieval/context enabled. I don’t have visibility into your exact UI state from here, but the usual check is: in the chat sidebar (or a Sources / Context / Attachments / Files panel), confirm you see HtM 1-7-26.docx listed as an active source for this conversation (not just “uploaded somewhere”). If there’s a toggle like “Use files in this chat”, “Include attachments in context”, or “Reference documents”, it needs to be on; if the doc shows as “uploaded” but not “in context,” the chat can’t search it.

If you tell me what you see in that panel (literally the labels/buttons), I can walk you to the exact setting. If your current chat mode simply doesn’t support retrieval, then uploads function more like storage (for the workspace) and the reliable method in-chat is still: paste the relevant excerpt and I’ll quote/match it from there.

There is no panel on the Conversation Context screen, so the doc isn’t an “active source”, but only stored? (And if it’s only stored, what’s the purpose?) So.. does this mean the only option is to copy/paste in chunks of old scene(s) I want Narrova to reference in a new scene, before I paste the outline for the new scene? Or should I upload the older scenes somewhere else? Or…? Thanks so much for your help.

2

Of course! Let’s see if we can get this squared away for you…

I’m sure you’ve read the docs on Context, but for those who haven’t, it’s here: Context - Narrova - Dramatica Documentation. There should likely be more specific documentation in regards to Uploading Documents but the idea is that you have three Context “holders” that you can work with, and turn on and off.

You have the Conversation Context - which is when you upload a Document straight into the Narrova Composer (that box with the + sign). When you upload Documents here, ONLY this Conversation has access to those Documents. You can add as many as you want, and remove them as you see best fit. You will see a Conversation icon pop up in the Composer window that you can click to see what files have been uploaded to the conversation.

Secondarily you have a Story Context. This shows up when you create a new Story and you can add Documents into the Story Context through the Context Manager (accessible through Stories nav at the top, or through the Story icon (an open book) in the Composer window). What’s nice about the Story Context is that you can upload multiple files to it, and then several different Conversations can have access to those documents–without you having to upload them every single time.

Lastly, you have the Storyform Context which essentially works the same as the Story Context, except that its attached to a Storyform. This means that in addition to any Documents you may have uploaded, the Storyform itself will be added as additional context to your work with Narrova.

If you don’t have a Storyform yet, I would definitely setup a Story, and then you can add whatever Documents you have to that Context. We also added a “Upload into Context” feature for your Conversations which is a way of skipping past the download file/re-upload file process so that you can simply click that, choose the Context to upload the conversation into and we go about adding it automatically as if it was file already downloaded.

With story and narrative development it’s very important that you maintain control over the context, over what is saved, and what is not, and so we don’t automatically add any Conversations into “memory”.

You will know when Narrova is accessing documents within Context when it reports back that it is “reading through your sources” - that’s an indication that it has specifically read through any documents added to a context. Narrova may also append “callouts” (footnote-like markings) when it references specific parts of sources, and then will list out the ones it referenced at the bottom of the response.

Let me know if that helps or if there is something else I might be missing that I can help out with.