Property intelligence

Deal packs become reviewable feasibility evidence.

Commercial deal packs turned into a validated, traceable, model-ready document pipeline — scaled from 20 to 100 tenants.

The challenge

Commercial deal packs, read and retyped by hand.

Align Property Partners works through commercial property material — information memoranda, financial packs, and technical due-diligence documents — where figures are read from each file and copied into a feasibility model. As deal volume grows, that manual extraction is slow, and a single mistyped value can sit unnoticed inside a model that drives a decision.

The work needed to be repeatable and, crucially, checkable: extraction that could be validated, traced back to its source, and scaled well beyond what hand-keying could handle.

What we built

A document-intelligence pipeline from source files to model-ready output.

  1. 01

    IM PDF

    Information-memorandum PDFs enter the pipeline as the source of truth, with PDF extraction pulling the text and tables out of varied layouts.

  2. 02

    Financial pack

    Spreadsheet financial packs are read through XLSX extraction, with tenancy-table detection isolating the schedules that matter.

  3. 03

    Extract

    Fields are extracted across documents, with LLM enrichment filling the gaps where layouts differ from one pack to the next.

  4. 04

    Validate

    A validation suite checks each field, surfaces missing values, and produces anomaly and missing-field reporting before anything moves downstream.

  5. 05

    Trace

    Every extracted value is linked back to its source through a traceability CSV, so the output is auditable rather than taken on faith.

  6. 06

    Report

    Templates are populated into a review-ready model — and a dedicated expansion harness took the model from a 20-tenant limit to a 100-tenant model.

Proof of work

Extraction that is validated, traceable, and built to scale.

Commercial deal packs now move through a single pipeline: source files in, extraction and validation in the middle, traceability and model-ready output at the end. Every figure is checked against its rules, anomalies and missing fields are surfaced for review, and each value traces back to the document it came from. A dedicated expansion harness proved the same pipeline scaling from a 20-tenant limit to a 100-tenant model.

Real deal-pack extraction pathwayValidation and anomaly reportingTenant-model expansion harness

Build details

The components and patterns behind the system — built the way production systems should be.

PDF extractionXLSX extractionTenancy table detectionLLM enrichmentValidation suiteAnomaly reportingMissing-field reportingTraceability CSVTemplate population100-tenant expansion harness

Bring us one workflow. We'll map the system underneath it.

Pick the process your team keeps doing by hand. We'll return its inputs, decision points, failure modes, and the automation layer that makes it operational — concrete, not conceptual.