Add AI-assisted input for the Namespace field in the Pinecone node (and the equivalent “collection/table/schema” selector in Supabase embeddings setups). Provide a small “Use AI” option (sparkle button / toggle) that can generate or validate the namespace from upstream context (e.g., current user, active enrollment, course/group name) and return a plain string to the node.
The idea is:
I run multi-tenant/vector workflows where one user can have multiple enrollments, each mapped to a different namespace. Today I must precompute the namespace with extra Code/LLM nodes and pass it via expressions—this adds boilerplate and failure points. Letting the node itself ask AI to derive namespace from the current item (e.g., $json.user.id, $json.inscripcion.activa, $json.groupName) would keep flows compact for both ingest and query paths.
I think it would be beneficial to add this because:
Reduces node count and complexity (less Glue/Code).
Lowers errors from manual string building and mismatched namespaces.
Makes dynamic routing (tenant → namespace) first-class in vector nodes.
Consistent DX across Pinecone and Supabase vector workflows.
Safer if paired with optional constraints (regex allowlist, max length, fallback to fixed value).
Any resources to support this?
Pinecone namespaces: Indexing overview - Pinecone Docs
Supabase + embeddings/pgvector patterns: AI & Vectors | Supabase Docs
n8n AI/LLM nodes (for parity of UX): https://docs.n8n.io/integrations/builtin/ai/
Are you willing to work on this?
Yes—happy to provide flow examples, test on self-hosted n8n, and help validate the UX (button vs. toggle, prompt template, and fallback/validation rules).