Prior implementation · January–February 2026
Built on hands-on conveyancing AI experience
In January–February 2026, MatterFirst's lead AI engineer led the implementation of a document extraction and generation tool for an Australian conveyancing firm.
The tool was deployed into a conveyancing practice to prepare contract reviews from contracts of sale, section 32/vendor statements and disclosure statements. It used the same underlying system that now underpins MatterFirst.
Explore MatterFirst Document AI- Field-level accuracy in the earlier implementation
- 98%
- Real documents in the earlier test set
- 100+
- Pages per document in the earlier test set
- 10–600
- Saved per contract review in the earlier implementation
- ~30 min
- Processing an average 100-page document in the earlier implementation
- ~89 sec
Testing on more than 100 real documents, ranging from 10 to 600 pages, measured 98% field-level extraction accuracy against the required outputs. The scoring included correctly leaving absent fields blank.
Team time tracking and practitioner feedback indicated approximately 30 minutes saved per contract review. The lead engineer reports approximately 89 seconds to process an average 100-page document; this is processing time, separate from lawyer review. These figures describe the earlier implementation and are reported by the lead engineer who carried out the work.