Generate sales emails based on business events with Explorium MCP & Claude

Explorium Event-Triggered Outreach

This n8n and agent-based workflow automates outbound prospecting by monitoring Explorium event data (e.g. product launches, new office opening, new investment and more), researching companies, identifying key contacts, and generating tailored sales emails leveraging the Explorium MCP server.

Template

Workflow Overview

Node 1: Webhook Trigger

Purpose: Listens for real-time product launch events pushed from Explorium's webhook system.

How it works:

• Explorium sends HTTP POST requests containing event data
• The webhook payload includes company name, business ID, domain, product name, and event type
• Pay attention: Product launch is just one example. You can easily enroll to many more meaningful events. to learn about events and how to enroll to events, visit the events documentation.

Node 2: Company Research Agent

Agent Type: Tools Agent

Purpose: Enrich company data after an event occurs.

How it works:

• Uses Explorium MCP via the MCP Client tool to gather additional company data
• Uses Anthropic Claude (Chat Model) to process and interpret company information for downstream personalization

Node 3: Employee Data Retrieval

Purpose: Retrieve prospect-level data for targeting.

How it works:

• Uses HTTP Request node to call Explorium's fetch_prospects endpoint
• Filters prospects by:
  • Company business_id
  • Departments: Product, R&D, etc...
  • Seniority levels: owner, cxo, vp, director, senior, manager, partner, etc...
  • Pay Attention: Follow our fetch prospect documentation for the full list of filter and best practice.
• Limits results to top 5 relevant employees
• Code nodes handle:
  • Filtering logic
  • Cleaning API response
  • Formatting data for downstream agents

Node 4: Conditional Branch - Prospect Data Check

If Node: Checks whether prospect data was successfully retrieved

Logic:

• If prospects found → personalized emails per person
• If no prospects → fallback to company-level general email

Node 5A: Email Writer #1 (No Prospect Data)

Agent Type: Tools Agent

Purpose: Write generic outbound email using only company-level research and event info.

Powered by: Anthropic Chat Model

Node 5B: Loop Over Prospects → Email Writer #2 (Personalized)

Agent Type: Tools Agent

Purpose: Write highly personalized email for each identified employee.

How it works:

• Loops through each individual prospect
• Passes company research + employee data to LLM agent
• Generates customized emails referencing:
  • Prospect's title & department
  • Product launch
  • Role-relevant Explorium value proposition

Node 6: Slack Notifications

Purpose: Posts completed emails to internal Slack channel for review or testing before final deployment.

Future State: Can be swapped with an email sequencing platform in production.

Setup Requirements

Explorium API Access

• MCP Client credentials for company enrichment and prospect fetching
• Registered webhook for event listening

Get explorium api key

n8n Configuration

• Secure environment variables for API keys & webhook secret
• Code nodes configured for JSON transformation, filtering & signature validation

Customization Options

Personalization Logic

• Update LLM prompt instructions to reflect ICP priorities
• Modify email templates based on role, department, or tenure logic
• Adjust fallback behavior when prospect data is unavailable

API Request Tuning

• Adjust page_size for number of prospects retrieved
• Fine-tune seniority and department filters to match evolving targeting

Future Expansion

• Swap Slack notifications for outbound email automation
• Integrate call task assignment directly into CRM
• Introduce engagement scoring feedback loop (opens, clicks, replies)

Troubleshooting Tips

• Validate webhook signature matching to prevent unauthorized requests
• Ensure correct business_id is passed to prospect fetching endpoint
• Confirm business enrichment returns sufficient data for company researcher agents
• Review agent LLM responses for correct output structure and parsing consistency