Skip to content
Regi Ellis

Client application

Harvest

Shared document processing, intake, and background services supporting DHN’s applications.

Built for
DHN Attorneys
Status
Deployed 2025
My role
Design & development

The need

Several applications depend on the same document and intake services.

DHN’s applications need to identify incoming documents, extract association-specific details, and track files as staff handle them. Those tasks require consistent rules and shared records across the firm’s software.

What I built

Shared services for the firm’s applications.

I built Harvest to provide reusable document processing and background services. Dyson uses its association profiles to classify documents and works with the intake queues Harvest maintains. The intake dashboard reads Harvest’s activity records to show handoffs and progress.

Harvest also handles attachment routing, notifications, and reply preparation. These services support the applications staff use each day, with shared document rules and a record of the work completed in the background.

Harvest / Shared application services

DocumentsFiles & association profiles
Intake activityMessages & staff actions

Harvest

Document processing, intake routing, and activity records

  • DysonClassification & intake queues
  • Intake dashboardHandoffs & progress
  • Staff emailNotices & reply drafts

Shared services support the applications staff use each day.

Implementation

How I approached the work.

Reusable document rules

Association profiles define how documents are identified and which fields to extract. Applications can use those rules without each maintaining a separate document reader.

Recorded handoffs and staff control

Background jobs record intake activity and check for messages already handled. Staff choose which messages receive an acknowledgment or an AI-assisted reply draft; substantive drafts remain available for review in Gmail.

Built with Technical details
  • Python + Click (CLI)
  • Gmail API + service account DWD
  • Pluggable pipeline framework
  • pdfplumber + PyPDF2 text extraction
  • 300+ HOA YAML extraction profiles
  • Claude API + systemd

Have a process
that needs better software?

Tell me what needs to change, your budget range, and when you’re ready to begin.

Discuss your project