Quick overview
This workflow runs two daily schedules to research crowdfunding campaign leads from Google Sheets using Firecrawl, DeepSeek/OpenAI, Jina AI, Perplexity, Hunter, and Lemlist, then verifies emails and scores each enriched lead with a 0–100 rating and priority label.
How it works
- Runs daily at 1am, reads unresearched rows from a Google Sheets “Leads” tab, and processes each product one at a time.
- Scrapes each campaign page with Firecrawl, cleans the markdown, and uses DeepSeek to extract key campaign details like amount raised, backers, and campaign dates.
- Uses a LangChain research agent (DeepSeek with OpenAI fallback) to research the company and decision maker, calling Jina AI, Perplexity, and Hunter as needed.
- Extracts structured contact signals (name, LinkedIn, domain) and enriches the person in Lemlist to find/verify email and phone, then waits and fetches the enrichment results.
- Combines research and Lemlist data into a fixed schema (converting non-USD amounts via a currency conversion API), verifies the email with QuickEmailVerification when present, and updates the matching row in Google Sheets.
- Aggregates the research run results and appends a summary row to the Google Sheets “Dashboard & Reports” tab.
- Runs daily at 5am, reads enriched leads from Google Sheets, scores each lead with DeepSeek into a 0–100 score and TOP/HIGH/MEDIUM/LOW priority, saves the results back to the sheet, and logs a scoring summary to “Dashboard & Reports”.
Setup
- Add credentials for Google Sheets, DeepSeek, OpenAI (fallback), Perplexity, and Lemlist.
- Replace YOUR_GOOGLE_SHEET_ID and ensure your spreadsheet has a “Leads” tab (with columns like Product Name, Product URL, Score, Priority) and a “Dashboard & Reports” tab for run logs.
- Set API keys for Firecrawl, Jina AI, Hunter, QuickEmailVerification, and the currency conversion API in the HTTP request/tool nodes.
- In the scoring prompt, replace the placeholders for your company name and business model to match how you want leads evaluated.