Recently, openai released beta v2 of their GPT assistant tech; as a result, file structuring has changed, theoretically for the better, where they have the vector store, which obviates a branch of development internally for the moment. Original specifications are found now in comments
[x] upload endpoint
[x] routes.mjs: member-route only /upload
[x] functions.mjs, api-functions.mjs: upload(ctx) function routes to avatar
storing in api-functions, with member validation wrapper
[ ] upload: internally-stored; 2 of 3 successive/additive states
[x] validity check (if exists, ignore/select, no overwrites at this moment)
[ ] define and create endpoint /upload
[x] file
[x] name
[x] context
[ ] content summary
[ ] other standard metadata
[x] state-1: vector-encode file
[ ] state-2: noSQL creation of asset in db
[x] state-3: store raw (out-of-scope of issue)
[ ] link: externally-saved ingestible (or even undigestible, I assume, would just continue to try to better consume itself)
[ ] ai-connection-assistant embed
[ ] assign/change purpose
[ ] execute purpose
[ ] suggest
produce a sensible question-blurb shown beneath its thumbnail; sensible here implying contextual awareness - i.e., what it has, what it doesn't have
from: What other categories should I have in order to best represent a jpg photo of blurbity?
to (after # maturation epochs or gpt-benched competency): what info, and formulate question thereof, would help expand category this understanding ${ send text } of blurbity-stuff
secure Assistant
protect all functionality behind known login
[ ] frontend interface
visual partially distinct from chatGPT
essentially has some additional icons, maybe is a sub-agent sitting inside of a parent? (Don't like it initially, give it another think)
full-screen concept with sub-widget architecture
'files' themselves capable of being in 3 states, not exclusive, but distinct
stored in noSql as object
each asset can have its own intelligent wrapper from which to query itself, based upon what skills 'core' has
vector concepts stored
raw file stored
Out-of-Scope
raw file storage
obviously this is a must-have, but foresee relationship with `permenant.org
Asset Assistant
Upload functionality
Recently, openai released beta v2 of their GPT assistant tech; as a result, file structuring has changed, theoretically for the better, where they have the vector store, which obviates a branch of development internally for the moment. Original specifications are found now in comments
routes.mjs
: member-route only/upload
functions.mjs
,api-functions.mjs
:upload(ctx)
function routes to avatarmylife-avatar.mjs
: instantiates Asset Assistantasset-assistant.mjs
: enhance/refactor for new vector-storemylife-factory.mjs
: stores vectorstoreId to datacoremylife-llm-services.mjs
: upload(), list() from openAI