Selected Systems & Automation Case Studies

I build practical internal tools for businesses with messy operational workflows, especially where teams are still copying information between messages, PDFs, spreadsheets, task lists, delivery routes, and accounting systems.

These case studies summarise selected systems I designed and built to solve real operational problems. They describe the problem, approach, tools, and transferable design thinking. Proprietary code, credentials, customer records, internal screenshots, and company-specific implementation details are not shared.

1. AI-Assisted Order Processing System

An AI-assisted workflow system designed to convert unstructured order messages into structured operational data.

The system processes natural-language order inputs and extracts key information such as product names, quantities, packaging units, pricing information, and delivery timing. The output is prepared for downstream workflow processing, with human review before final submission.

The focus of the project was not simply to “use AI”, but to create a practical operational layer that could reduce manual interpretation, improve consistency, and support staff working with messy real-world order data.

Key areas: LLM-assisted data extraction, human-in-the-loop validation, product matching, data transformation, workflow automation, and operational usability.

2. ClickUp to Xero Invoice Export Pipeline

A workflow-to-finance automation pipeline designed to convert completed operational task data into invoice-ready exports.

The system extracts structured order and item data from task-management records, validates key fields, and formats the output into CSV files suitable for financial processing.

The project created a cleaner bridge between operational workflows and invoicing, reducing repetitive manual reconciliation work and improving consistency between production records and finance administration.

Key areas: task data extraction, CSV automation, workflow-to-finance integration, invoice preparation, operational reconciliation, and process automation.

3. PDF Packing Slip Order Ingestion Tool

A PDF order-ingestion tool designed to extract structured order data from external packing slips and prepare it for workflow submission.

The system extracts customer details, order numbers, dates, product lines, SKUs, and quantities from PDF documents. It uses deterministic parsing where the document structure is predictable, with AI fallback logic where the structure is incomplete or inconsistent.

The project reduced the need for manual re-entry of external order documents and created a more reliable ingestion layer between external order formats and internal operational workflows.

Key areas: PDF parsing, structured data extraction, AI fallback logic, workflow ingestion, data validation, and operational automation.

4. Route Optimisation & Delivery Planning Tool

A delivery-planning tool designed to support route optimisation using stored customer location data and external routing logic.

The system allows users to select delivery points, retrieve location data, group stops, and generate optimised delivery sequences with estimated distance and duration.

The project focused on reducing manual route-planning effort and improving the usability of delivery coordination workflows.

Key areas: route optimisation, location data handling, logistics planning, API integration, delivery coordination, and internal tooling.

5. OpenStreetMap Lead Intelligence Pipeline

A geospatial lead intelligence pipeline built using public OpenStreetMap data to identify potential business leads along selected delivery routes.

The system processes raw map data, extracts relevant business locations, calculates distance from target routes using coordinate projection, filters duplicates and unwanted categories, and exports structured lead data.

The output can include business names, categories, addresses, contact details, websites, opening hours, coordinates, and distance from the selected route.

The project demonstrates how open geospatial data can support low-cost business development, route-based prospecting, and operational planning.

Key areas: Python, OpenStreetMap, geospatial analysis, data extraction, lead generation, CSV automation, route-based filtering, and business intelligence.

Footer note

Note: These case studies describe architecture, workflow design, and transferable problem-solving approaches. They do not include proprietary code, credentials, customer records, internal screenshots, or company-specific implementation details.

View related LinkedIn projects: Projects | Cobus (Jacobus Francois) Mostert | LinkedIn