The work, and what it did.

The problem · Prospecting was manual: designing a medal mockup for each lead by hand, then writing a personalised email — slow, and one wrong year or misspelled club name killed the pitch.

What we built · A 14-node n8n “Cadillac Prospecting System.” Google Sheets feeds prospects into image generation (two base medal designs each), deterministic text-and-logo overlay via Cloudinary with a computed award year, a personalised HTML email, and an Outlook draft created through the Microsoft Graph API — with a full audit log written back to Sheets. Runs itself at 2 AM daily.

The problem · Building and maintaining exam content by hand didn’t scale — the academy needed an AI-driven test bank, weighted study priorities, and licensing content that could stretch across multiple states.

What we built · An AI exam-content system built over ~40 working sessions: glossary automation, weighted term prioritisation, AI-tagged and validated test-bank items, multi-state licensing content, and a modular, adaptive-learning textbook structure.

The problem · Customer-service response was slow and there was no after-hours coverage — calls went unanswered outside business hours.

What we built · A voice agent handling call routing, knowledge retrieval, and a structured decision tree, live around the clock.

The problem · The company wanted a no-code product that could turn a simple form into a branded, downloadable compliance document (RAMS and CPP, to CDM 2015).

What we built · A document-generation pipeline: a Tally form webhook feeds a field-mapper and a RAMS/CPP router, an LLM drafts the content, and the result is rendered to a branded PDF and delivered by email.

The problem · Everything was manual — inbound leads (70–300 a month across several channels), invoicing, client comms, and logistics for a 20-plus person team.

What we built · An n8n CRM: inbound leads captured from Gmail into a Google Sheets database, with instant smart-routed auto-replies and Slack alerts to the right person.