Land sourcing extended into Scotland in 13.3 days

01The challenge

Verafind's searches covered England and Wales. Scotland has its own legal system and its own land register, ScotLis, and the owners' contact details sit in the prose of each title sheet rather than in fields a system can read. Northern Ireland adds a third register, built around a Folio Number that holds each title's description, ownership and encumbrances.

02What we did

1. 01Registry access — registered with ScotLis and obtained a sample of titles to test whether AI extraction would work
2. 02Acquisition — a programmatic inbox receives ScotLis purchase emails and pushes each title PDF into Amazon S3
3. 03Extraction — AWS Textract converts each PDF, then iterated GPT-4 Turbo prompts pull the owners' contact details out of the title prose
4. 04Enrichment and import — external data sources assess each parcel's development potential, and a translation layer converts it to Verafind's standardised land parcel format
5. 05Northern Ireland — the same approach adapted to the Land Registry of Northern Ireland, its compulsory registration and its Folio Numbers

03What we built

• ScotLis title purchases delivered straight to Amazon S3
• AWS Textract parsing of title PDFs
• GPT-4 Turbo extraction of owner contact details
• Development-potential enrichment from external data
• Translation into the standardised land parcel format
• Folio Number handling for Northern Ireland titles

04The result

Scottish titles now import into the main system in the same format as every other parcel, so Verafind can find and work opportunities in Scotland for Electric Land. Northern Ireland followed on the same pattern for similar effort, and the pipeline serves Verafind's other clients too. Electric Land went on to consider the Netherlands, and a plan for working with the Dutch Kadaster followed.