Skip to main content

​
PropTech Variance Commentary Tool

​
Overview

LostMind AI’s PropTech Variance Commentary Tool converts raw portfolio ledgers into CFO-ready variance narratives. It slots into the broader PropTech automation stack, sharing AI orchestration, security, and observability patterns with the SaaS platform.

​
Key Capabilities

  • Workbook Sanitisation — Redacts Knight Frank source files before any AI processing
  • Variance Detection — Highlights period deltas across thousands of GL and AP records
  • AI Commentary — Generates auditor-friendly explanations with configurable tone
  • Portfolio Views — Groups insights by property, manager, or account class
  • Compliance Guardrails — Ensures Australian data-sovereignty requirements are respected

​
Quick Start

Use the upstream repository scripts as a guide when operating locally: 
# 1. Install dependencies
pip install -r requirements.txt

# 2. Sanitise source workbook (mandatory)
python sanitize_data.py

# 3. Build the SQL model
python excel_to_sql.py

# 4. Generate commentary draft
python scripts/run_variance_commentary.py --period 2024-12
AI features connect through the shared ai_integration/unified_client.py module. Configure provider access in .env and route sensitive data to on-prem Gemma if compliance demands it.

​
Documentation Snapshot

Historical implementation notes, data dictionaries, and operating guides are stored in the _archived/projects tree for this initiative. Consult those files for detailed workflows while the live docs are pruned.

​
Support

Raise issues through the LostMind AI platform team or reach out via the internal support channel when variance commentary output needs review. Content derived from PropTech finance tooling requirements and internal design notes.