An n8n template combines an Ollama agent, three Supabase vector knowledge bases, Postgres memory, request validation, deduplication, and confidence-based escalation across Slack, Gmail, and callbacks.
The template indexes provided learning documents in MongoDB Atlas and uses a Gemini Telegram agent to retrieve relevant passages before answering questions.