n8n Support Automation: Ollama RAG + Supabase KBs to Slack/Gmail

Turn support tickets into instant, source-backed replies—using Ollama RAG + Supabase, delivered to Slack or Gmail

This n8n support automation workflow ingests incoming support requests via webhook, validates and rate-limits them in Supabase, then uses an Ollama-powered LangChain agent with three Supabase vector knowledge bases (and optional Wikipedia search) to generate answers. High/medium-confidence responses go to Slack, Gmail, or an HTTP callback; low-confidence or human handovers are escalated to operators.

What this workflow does

• Receives webhook support requests (POST) and validates the x-api-key header—returns 401 if missing/invalid.
• Normalizes and logs requests in Supabase to enforce per-session rate limiting, returning 429 when limits are exceeded.
• Rejects invalid/duplicate messages: checks for an empty body, enforces a 2000-character limit, and deduplicates by message_id in Supabase.
• Returns 202 Accepted immediately while it processes: sanitizes customer queries to mask common PII patterns and detects explicit requests to speak with a human.
• Runs a LangChain agent on Ollama with Postgres chat memory, retrieving context from three Supabase vector KBs (and using Wikipedia when needed).
• Routes outputs based on the agent’s JSON confidence: sends replies to Slack, Gmail, or an HTTP callback, and writes a scrubbed audit log with sources and confidence to Supabase.
• Escalates when confidence is low or a human is requested—queues escalation in Supabase and alerts operators in Slack during business hours (emails on-call off-hours).

Use cases

• Automated L1 support for SaaS: answer FAQs with internal KBs and citations.
• Prevent support overload: rate-limit and deduplicate inbound messages.
• Smart routing: send confirmed answers to Slack/Gmail while escalating uncertain cases.

Technical details

• Integrations/Nodes: if, set, code, Gmail, Slack, switch
• AI layer: Ollama + LangChain agent, Postgres chat memory, retrieval from 3 Supabase vector knowledge bases, optional Wikipedia
• Data control: Supabase request logging, rate limiting, deduplication, and audit/escalation queues