This process calls the JWT-protected Supabase ingestion function. The function itself fetches only its allow-listed official PDF URLs; this browser does not upload or fabricate ESG content.
Keep this tab open until completion. Reports are processed in small batches to stay within serverless execution limits. Re-running is safe: already-ingested page batches are skipped.
What gets created
Source chunksPage-aware text from official PDFs
384D embeddingsSupabase gte-small, mean pooled and normalized
Vector indexPostgreSQL pgvector HNSW
Keyword indexGenerated tsvector + GIN
Source citationsDocument title, URL and page retained