Building a lead-scoring pipeline in n8n — deliberately without an AI agent
FlakyBeyond5850 · reddit · 2026-09-23
The author built a production-style pipeline in n8n to research and qualify marketing agencies in Bengaluru, deliberately avoiding an AI agent.
Workflow
- Accepts a niche, location, and company limit
- Searches with Tavily, filters out directories and duplicate domains
- Fetches and cleans site content, falling back to ScrapeGraphAI when normal extraction fails
- Structured lead analysis via OpenRouter
- 0–100 scoring stored in Google Sheets, Markdown report generation, and a separate error-handler workflow
Core design decision: the LLM only analyzes companies; score boundaries and qualification rules are enforced in JavaScript (80–100 high, 60–79 medium, 0–59 low). The rationale: lead scoring must be predictable and auditable, while an agent introduces unnecessary tool-selection and execution variability. The author invites debate on whether this workflow should use an agent at all.
More from coding & agent
- Stripe launches WebMCP endpoint letting AI agents complete full web purchases — jeff_weinstein · 2026-09-23
- What would an ideal plugin editing environment inside ChatGPT/Codex look like? — Former_Worldliness70 · 2026-09-23
- boxd_sh praised as a major unlock for AI agents needing persistent computers — tobowers · 2026-09-23
- Open-source AI intraday trading bot ranks Nifty 50 every 15s and trades via Zerodha Kite — IndraVahan · 2026-09-23
- Debugging agent planners: which state do you actually save to reproduce a bad plan? — fishyguy3123 · 2026-09-23
- Yoav Goldberg: some tasks just need deterministic rules — agents can write them — yoavgo · 2026-09-23