Workflow Automation

Connecting systems that don't talk to each other. CRM events, email parsing, document routing, notifications, validation, error handling, and reporting - automated pipelines replacing manual coordination.

This capability has been applied across 12 projects spanning automotive, hospitality, financial services, heavy equipment, and AI platform development.

Automotive - Customer Delivery Notifications

A dealership group needed customers and staff notified at each stage of the vehicle delivery process. The existing workflow was manual - someone monitored incoming emails from the dealer management system and sent status updates by hand.

The automated system monitors CRM delivery events, validates incoming data, and sends formatted notifications to customers and internal staff at each milestone. Includes test modes for staging, daily volume reporting, and error handling that alerts the team when something fails silently.

Automotive / Entertainment - Voice Agent Notification Pipelines

Two production voice agents across different industries produce structured data from phone conversations. That data needs to reach the right people in the right format - booking confirmations to service teams, accessibility requests to venue staff, after-hours summaries to managers.

A shared pipeline handles webhook ingestion, data formatting, recipient routing, and delivery across email and CRM. The same architecture serves both agents despite different data schemas and delivery requirements.

Automotive - Performance Reporting Pipeline

A voice agent analytics system pulls execution history from two separate automation platforms (the result of an infrastructure migration), merges data chronologically, and produces weekly performance reports. Data consolidation across the migration boundary required careful handling of schema differences between platform versions.

Automotive - Deal Pack Document Validation

Vehicle deal packs contain dozens of documents in inconsistent PDF formats. Staff manually reviewed each pack for completeness.

A document extraction pipeline reads deal pack PDFs, identifies document types, pulls key fields, and flags missing items. The extraction feeds into a validation workflow that checks completeness against expected document sets for each deal type.

Financial Services - Invoice Processing

An accounts receivable workflow that processes incoming invoices, extracts structured data, matches against expected records, and produces client-facing PDF summaries. Handles format variation across suppliers without manual intervention.

Automotive - DMS and CRM Data Pipelines

Two data integration workstreams for a dealership group. The first extracts data from the dealer management system data warehouse via a containerised ETL pipeline with schema querying, row count validation, and scheduling. The second integrates with a CRM's REST API via OAuth to evaluate AI-assisted customer engagement performance across 1,100+ contacts.

Heavy Equipment - Form Delivery Automation

RPA automation for delivering forms to customers at a heavy equipment dealership. UiPath process with a dispatcher pattern separating job orchestration from execution. Dual-environment configuration (production and UAT), exception handling with screenshot capture, and email error reporting. Multiple release cycles across the main process and dispatcher.

Hospitality - Document Extraction Routing

The automation layer connecting a document extraction system to operational workflows. Incoming extraction outputs are parsed, validated against expected schemas, and routed downstream. Built as the integration bridge between AI extraction and business processes.

Internal - Infrastructure Management

Production infrastructure management across client and internal environments. Self-hosted automation platform deployment, client VM management (one sitting behind a client VPN requiring specific networking), and operational responsibility for keeping production services running.

AI Platform - Workflow Orchestration (214 commits)

Core contributor to a workflow orchestration platform at an AI startup. 214 commits to the managed-workflows system that powers automated pipelines across the platform. Separate work on ingestion pipelines including website connectors and data scraping infrastructure.

The Pattern

Every project follows the same structure:

1. Trigger - a webhook, email, CRM event, API call, or scheduled job initiates the workflow
2. Extract - structured data is pulled from an unstructured source (email, PDF, transcript, API response)
3. Validate - extracted data is checked against expected formats, flagged for exceptions
4. Route - validated data is sent to the right system or person in the right format
5. Report - automated reporting on volume, errors, and exceptions

The technology varies - n8n, UiPath, Power Automate, custom Python pipelines, platform-level orchestration - but the pattern is consistent.

Stack

n8n, UiPath, Power Automate, Python, Docker, AWS Textract, CRM and DMS APIs, webhook-driven architectures.