Hiring Automation Burn Charts Cosmic Careers Animated Construction Exoplanets Blueprints to Blastoff Cosmic Cookbook IBACOS Standards Penn Forest Map

Case Study: Hiring and Transfer Automation

Transfer Selection form listing nearby Coen Markets stores with distance estimates and toggle switches — Figure 1. Transfer Selection form showing nearby Coen Markets stores with per-store distance in miles and minutes, plus toggle selection.

Lyon Ritchie built and documented business automation for Proxigee Services' client Coen Markets: two n8n agent systems handling interview scheduling and employee transfer selection, plus a shutdown documentation package that keeps every workflow readable and portable. Proxigee Hiring and Transfer Automation ran for nine months, produced 40 backed-up workflows, and spun off a published open-source tool – an n8n community node for Geohash operations.

The Brief

Coen Markets runs hiring and employee transfers across systems that do not talk to each other: UKG for HR records, Airtable for tracking, Twilio for candidate messaging, and n8n for orchestration. Two agent systems carried the load: an Interview Agent (scheduling assistant, calendar proxy, UKG hiring updates, disposition tracking) and a Transfer Agent (opt-in forms, transfer-set assignment, proximity matching so employees see stores within 30 miles of home, each with per-store distance estimates).

Unlike our typical 3D and animation engagements, this project was pure business automation – much of it built with AI coding agents – and it ended with a shutdown deliverable: the client needed every workflow, schema, and automation captured as documentation that survives without the original SaaS accounts.

Our Approach

We developed the n8n workflows with AI agents in the loop – Gemini for workflow construction, Antigravity with n8n-MCP, and Hermes with a browser harness to extract Airtable automation definitions that Airtable itself will not export. Agent output moved fast but needed guardrails: workflow backups went to our private code repositories (sidestepping the client’s 5-backup account limit), and agent-generated documentation almost always required a manual fixing pass, highly dependent on prompt, model, and context limits.

For proximity matching, n8n had no native GIS/spatial operations, so we built one: the n8n-nodes-geohash community node (encode, decode, bounds, adjacent, neighbours), published to npm, with a Code-node-only alternative workflow for cloud and restricted instances that cannot install community nodes. Read more in our lab note on the Geohash node.

Public-facing forms were designed in PenPot and styled with design tokens captured from the client’s live Coen Markets site, so the transfer-selection experience matches their brand.

The Result

Lyon Ritchie delivered a closedown package containing the full n8n workflow library (40 workflow, automation, and schema JSON files across Interview and Transfer agents), rendered HTML documentation with agent overviews, per-workflow reference, and schedule/credential mapping, plus Airtable schemas and automation descriptions – published to the client’s review site and versioned code releases.

The engagement ran December 2025 to September 2026. Along the way it produced a reusable open-source artifact (the Geohash node, now at v1.0.6) and a repeatable pattern for AI-assisted automation work: agents build, version control preserves, humans verify.

Posted on 2026-10-14 by Gary Ritchie.