Agreements and standard letters
were being prepared by hand.

The Problem

The workflow relied on copying previous documents.

For each new conveyancing matter (a property sale, a tenancy, a financing transaction), the firm's clerks would open a previous document, strip out the client details, and re-draft it for the new matter. SPA agreements, tenancy contracts and 15 client-letter types were assembled manually from previous files.

The process was slow, error-prone, and entirely dependent on the clerk remembering which document to use as the base. Mistakes in property details, party names, or land title references meant hours of correction before anything could go to the client.

In a jurisdiction with specific local legal requirements, the inconsistency between document versions was also a compliance risk.

The Build

Client data in. Formatted legal document out.

The system is a set of Python generators, one per document type. Each generator takes a structured input (party names, property details, transaction values, dates, land title references) and produces a correctly formatted, court-ready document using a locked template.

Templates are maintained as versioned master documents, separate from the generation logic. When a clause needs updating (a standard fee change, a new statutory requirement), the template is updated once, and future documents generated from that template reflect the update. The wording no longer depends on which base file a clerk happened to copy.

The SPA generator handles the most complex case: title type, encumbrances, special conditions, stamp duty calculations, and completion timeline, all populated from the input and cross-checked against the land details before output.

The Result

The configured matter suite from one filled-in spreadsheet.

The clerk enters the client data once and runs the generator. The system produces the documents for that matter together: SPA, tenancy agreement, requisition letters, completion notice, and any of the 15 standard letter types the matter requires, as finished Word files.

The generation logic enforces completeness: if a required field is missing, the system flags it before generating rather than producing a document with a blank where party details should be.

The system is in production on the firm's matters, and the compliance risk from document version drift is designed out: the wording lives in one versioned place rather than in whichever file a clerk copied last.

Stack

What it was built with.