AI Job Intake Automation for a Canadian Recruitment Agency

Overview

A Canadian recruitment agency was retyping every new job order into Bullhorn by hand. I built an intake pipeline where a Claude agent reads each request, from an email or the intake form, pulls out the job details, checks them against Bullhorn and creates the record. The right recruiter gets an alert and the client follow up goes out on its own, so new roles get a response 40% faster.

About the client

A recruitment agency in Canada with 50 staff. The team runs its desk on Bullhorn and holds an internal database of about 500,000 candidate records built up over years of placements.

The Problem

What they were dealing with

New job orders arrived by email and through an intake form, in whatever shape the client chose. Someone had to read each one, work out the role, location, rate and requirements, and type it all into Bullhorn before a recruiter could start. On busy days new roles waited in the inbox, and following up with the client depended on who remembered.

Before

The Solution

How I solved it

The goal was simple: a new job order should reach Bullhorn correctly without anyone retyping it, and the client should hear back fast.

An n8n workflow running with APIs on AWS picks up every request from email and the intake form. A Claude agent built with the Claude Agent SDK reads the request and any attachments and returns the job details in a fixed structure. A validation step checks each field against Bullhorn before anything is written, and when it cannot match a value with confidence it stops and asks a person instead of guessing. Once the job exists in Bullhorn, the right recruiter is alerted and the client gets a prompt acknowledgement and follow up.

The Result

What changed