The Problem
A job-search campaign across the UK and Ireland meant checking hundreds of careers pages and writing every email and follow-up by hand.
What I Built
A Python multi-agent pipeline that automates the entire outreach sequence:
Company research: Reads target contacts from Google Sheets, detects each company's applicant-tracking system (Greenhouse, Lever, Workday and others) and pulls live roles through their APIs.
AI email writing: Claude picks the best-matching role and writes a personalised email for each contact. Every email is specific to what that company is actually hiring for.
Multi-inbox sending: Sends through three Gmail accounts within UK business hours, respecting daily limits and each country's bank holidays.
Reply classification: One LLM call sorts every reply into a real answer, an out-of-office or a "no longer here" notice, and the sheet updates itself.
Three-step sequences: Every contact gets a researched, relevant email and at most two follow-ups. Sequences close themselves when nobody replies.
The Outcome
End-to-end automation from company research to follow-up, with no manual tracking. Each contact gets a personalised, role-specific email. Runs on a schedule.
0 likes · 2 views · Nov 15, 2026
Python multi-agent pipeline that researches companies, writes personalised emails with Claude and runs three-step sequences across multiple inboxes.
Timeline
Jan 1, 2026 - Ongoing