Melbourne Property Intelligence Dashboard

melbourne-property-intelligence-dashboard

A property management workspace for handling email, property records, and follow-up tasks, with AI analysis presented for human review.

Role: Developer

Workflow illustrations

Reconstructed views using fictional data to explain the workflow. These are illustrations, not screenshots of the live application.

Illustrative mailbox view showing a fictional maintenance email and an AI suggestion awaiting review. — Mailbox · read an email alongside a suggestion for review.

Illustrative property view showing fictional property Example A and a linked maintenance task. — Property context · connect a property record with follow-up work.

Illustrative task list showing fictional maintenance follow-ups and their statuses. — Task follow-up · make outstanding work and status visible.

Highlights

ReactTypeScriptViteNode.jsExpressPostgreSQLPrismapg-boss

Connected property records, mailbox activity, and follow-up tasks through a React interface and an Express REST API.
Implemented IMAP synchronisation, SMTP replies, attachment handling, and drafts.
Separated interactive requests, mailbox jobs, and AI analysis so long-running work has its own processing path.
Added team permissions, mailbox authorisation, AI budget checks, and data retention controls.

Problem

Property follow-up involves moving between messages, property records, and outstanding tasks. The workspace needs to preserve that context without making users reconstruct it for each action.

Private correspondence also needs clear access rules and a distinction between model suggestions and user decisions.

Approach

Connect email, properties, and tasks through shared records, while keeping replies and decisions under user control.

Give interactive requests, mailbox processing, and AI analysis separate responsibilities so a slow background operation does not define the request flow.

Result

The workspace brings property context, mailbox handling, tasks, and reviewable AI suggestions into one application. The public website provides an entry point to the project.

Implementation

Property and task workflows

Built React and TypeScript views for properties, tasks, email, and team administration.
Used Express, Prisma, and PostgreSQL for application data and applied team permissions to business actions.

Mailbox processing

Integrated IMAP synchronisation and SMTP sending with message bodies, attachments, drafts, and manual replies.
Used PostgreSQL-backed job queues for mailbox work outside interactive API requests.

AI review and data controls

Added structured email analysis and summaries as suggestions for review.
Separated AI processing from mailbox access and applied model-request budget checks.
Implemented agreement, mailbox authorisation, retention, and cleanup workflows.

Lessons

A completed request, a completed mailbox job, and a completed AI analysis are separate outcomes and need separate checks.
When handling private correspondence, automation boundaries and visible review steps are part of the workflow design.