Build the Pipeline

Pillar 2: Build the pipeline (quick-start)

This is where you build the machine that fills your pipeline: an n8n flow that pulls jobs straight from company career pages, searches the whole market by keyword, flags every role where you already know someone, and drops it all into your tracker on a schedule. Written for someone who has never opened n8n. Use the attached Part 1 and Part 2 docs, the two code files, and the troubleshooting prompt as you build.

The core idea before you start: you stop scrolling job boards. You build a system that hands you a sheet of relevant jobs, already tagged by whether you have a warm path in.

n8n in plain English

n8n is a visual automation tool. You drag "nodes" (steps) onto a canvas and connect them into a "workflow," and data flows through as "items." You'll use four node types: Trigger (starts it), Google Sheets (reads and writes your sheet), HTTP Request (fetches from a URL), and Code (reshapes data). You paste our scripts into the Code nodes, so you never write code yourself. Sign up for a free n8n Cloud trial, nothing to install.

Part 1: Tier A, pull jobs from company career pages

Your target companies go in, n8n hits the public job feed behind each company's careers page, clean job rows come out. No scraping, no login, the freshest possible source.

• Input: a Targets sheet with Company, ATS, Slug. Read the ATS and slug right off the careers URL, for example boards.greenhouse.io/acme means ATS greenhouse, slug acme.

• The five-node flow: Manual Trigger, then Google Sheets (Targets), then Build Endpoint (Code), then HTTP Request, then Normalize (Code), then Google Sheets (Jobs).

• Paste the two Code node scripts from the Tier A code file. Start with one Greenhouse company to get a win, then add more. Greenhouse, Lever, and Ashby are confirmed; the others carry quick verify notes.

Part 2: Tier B, Tier C, dedupe, and schedule

• Tier B, market keyword search: one search across Hiring.cafe via Apify catches roles at companies not on your list. Get a free Apify token, then Search Criteria, then HTTP POST, then Normalize.

• Merge and dedupe: combine Tier A and Tier B with a Merge (Append) node, then a Dedupe Code node drops repeats.

• Tier C, warm leads: paste your LinkedIn connections export into a Connections tab, add a Get Connections node, then a Match Connections Code node tags every job Warm or Cold and records who you know there.

• Write it out tagged, then swap the Manual Trigger for a Schedule Trigger and activate the workflow so it runs itself.

The whole machine

Schedule Trigger to [Tier A: Targets, Build Endpoint, HTTP, Normalize] plus [Tier B: Search Criteria, HTTP POST, Normalize], then Merge, then Dedupe, then Match Connections, then your Jobs sheet.

Troubleshooting

n8n is finicky when you're new, so grab the attached troubleshooting prompt. Paste it into any AI chat along with your error and the failing node's input and output, and it will walk you to the fix. Most issues are one of a few things: a Code node set to the wrong Mode, an expression field not switched on, a wrong slug, or a field named differently by that ATS.

What you have

A running pipeline that sources jobs from your targets and the whole market, dedupes them, flags where you have a warm path, and refreshes on a schedule into your tracker. From here it's pure execution: Pillar 3 works your warm leads, Pillar 4 your cold leads.

Go deeper: the docs, the code, and the prompt

This is the quick-start. The attached Part 1 and Part 2 docs walk every node click by click, the two code files hold the paste-ready scripts, and the troubleshooting prompt is there for when a node misbehaves. Bring your questions to the group calls.