Recipes & example workflows

The package ships ready-to-import workflows in its examples/ folder. Open n8n, choose Import from File (or paste the JSON), set your SocialMate credential, and run. Browse them on GitHub:

AI concierge with full memory

Pro real-estate-deepseek-agent.json — a WhatsApp concierge that qualifies leads with the entire conversation in context.

1. SocialMate Trigger → message.received.
2. Message → Get AI Context → format: transcript, order: newest.
3. DeepSeek agent (any LLM node) with the transcript in its system message.
4. Message → Send Text → the model's reply to ={{ $json.data.chatId }}.

This is the pattern that makes SocialMate special — see AI agents with real memory for why Get AI Context beats the n8n memory node.

Auto-reply round-trip

Free auto-reply-trigger.json — a two-node workflow that acknowledges every inbound message. The simplest possible build, and it runs on Free.

1. SocialMate Trigger → message.received.
2. Message → Send Text → Chat ID ={{ $json.data.chatId }}, your reply text.

Scheduled batch send

Pro scheduled-bulk-send.json — a daily batch of personalised messages to people already expecting them (appointment reminders, order updates), sourced from a spreadsheet or database. One individual message per person, not a broadcast.

1. Schedule Trigger → daily at your chosen time.
2. (Optional) Google Sheets / Postgres → fetch today's rows.
3. Queue → Queue a Batch → a template with {{placeholders}} + the rows; the smart queue paces every send through anti-ban. Personalise every row — identical text to many contacts is blocked by the duplicate-content guard.

The rest of the library — all ready to import

Ten example workflows ship with the node, in its examples/ folder. Alongside the three above:

• AI agent that behaves like a human Proai-agent-tool.json — SocialMate wired in as an AI-Agent tool (usableAsTool), pre-loaded with the seed prompt: read the thread, mark it read, show typing, then reply.
• Expose WhatsApp to Claude / Cursormcp-server-trigger.json — build an MCP server inside n8n so an MCP client can drive WhatsApp, without running the standalone MCP server.
• Order confirmation from your storeorder-confirmation-webhook.json — a WooCommerce/Shopify "order created" webhook → Send Text. No WABA, no per-message fee.
• Drip onboarding Prodrip-onboarding-sequence.json — one trigger enqueues three scheduled sends, so a welcome sequence lands over days instead of all at once.
• Keyword routerkeyword-router.json — branch an inbound message on keywords and answer differently. The no-LLM way to deflect FAQs.
• New-lead alertnew-lead-alert.json — a first-time sender fires a notification to your team.
• Vision memory Provision-memory-loop.json — describe a photo or voice note with your own model once, cache it with Agent Memory, and never pay to analyse it again.