I shipped a live hiring pipeline — not a demo.

At Seisma I personally built a production n8n workflow on AWS that ran recruiting through to onboarding: Salesforce records, candidate self-service, documents, approvals, provisioning, and comms.

Observed, January–May 2026 — not an audited study: about 8-10 hires a month; about 60% less manual work; about 40% less account-management data entry.

Also built: a small HR assistant, locked to authorised people. Not a public chatbot. Not a platform.

LinkedIn

jasonwu.ai

What was broken

Hiring and onboarding were a pile of hand-offs: intake, records, documents, approvals, provisioning, and “please complete this” mail. People copied. Things stalled.

What I built

An end-to-end n8n workflow on AWS EC2, tied to Salesforce. Structured sales intake, candidate self-service, document generation, approvals, provisioning triggers, onboarding communications. I built it. It ran live.

What humans still did

Judgement stayed with people. The workflow moved the repetitive path. It was not an unsupervised agent fleet. After the first four or five hires it settled into about 8-10 hires a month (Jan–May 2026).

What this is not

Not a public HR bot. Not a product you can click. Not a platform hire. Not an audited ROI study. The numbers above are observed operating measures from that window.