Case Studies
AI agent systems in production
Each case study documents an AI-enabled operating system: the business risk identified, the agent workflow deployed, the human controls kept in place, and the commercial work it moved forward. These are not demos. They are production proof.
Native Booking Flow
Paid Scheduling Link → Owned Conversion Workflow
Replaced a paid scheduling dependency with a first-party booking flow backed by Google Calendar, preserving the booking outcome while reducing tool overhead and external handoff risk.
Dependency
Removed
Backend
Google Calendar
Friction
Minimal
AI Growth Reporting Agent
Autonomous Analysis → Delivery Backlog
Deployed an agentic automation that analyses ecommerce performance, writes the monthly growth report, drafts client comms, and creates delivery-ready recommendations in the task backlog.
Mode
Autonomous
Engine
AI Agent
Outputs
PDF + Tasks
AI Backlog Execution Agent
Part 2 · Autonomous Triage → Execution → Review
Extends the reporting agent into an autonomous execution layer: triaging backlog tasks, preparing implementation work, routing judgment calls, and closing the loop.
Mode
Autonomous
Engine
AI Agent
Control
Human review
Crate Clothing
Liquid → Headless Revenue System
Replaced theme-bound commerce logic with a structured, measurable revenue system. Structured merchandising, consent-gated analytics, and SSR performance installed as continuous infrastructure.
LCP Delta
31.9s → 2.2s
LCP Reduction
−93%
Engagement
Phase 1 Complete
Littl' Juey
3 Stores → 1 Multi-Market Platform
Consolidated three regional stores (NZ/AU/UK) into one multi-market Shopify setup. Used agentic migration workflows to adapt regional terminology while cutting platform and admin overhead.
Stores
3 → 1
Overhead
Consolidated
Localisation
Agentic