steps.txt

#give yourself more patiences

1. explore metadata, check each keys

2. define retriever
supabase?
relational database?, embeddings, content, id, ...
create a project, and a table + columns first emm...
https://supabase.com/dashboard/project/ohzwldyjckkuzbybaixs/editor/17248
enable vector in extensions under database

create table public.documents (
  id bigint generated by default as identity primary key,
  content text,
  metadata json,
  embedding vector(768),
  similarity float
);

create index for embedding!!!

add functions, advanced settings, sql language

create index on documents using hnsw (embedding vector_ip_ops);
alter table documents enable row level security;
create function match_documents_langchain (
  query_embedding vector (768)
)
returns setof documents
language plpgsql
as $$
begin
  return query
  select *
  from documents
  order by documents.embedding <#> query_embedding
  limit 1;
end;
$$;

3. define agent

4. define gradio