The one judgement in conveyancing you can’t hand to an agent
Summary
The article argues that while agentic AI can help with administrative tasks in conveyancing, it should not be trusted to make liability-bearing legal judgements without explicit, expert-authored rules and verified data. It describes Moverly's DiligenceEngine as a system designed to assess conveyancing risks using structured playbooks and provenance-tagged information, with an initial focus on auction property packs.
Why it matters
Residential property surveyors may encounter transactions where AI-assisted conveyancing influences risk assessment, document review and client expectations. The piece highlights the importance of verified data, clear accountability and cautious use of AI in areas that can affect legal risk and completion outcomes.
Key points
- Agentic AI is presented as useful for routine conveyancing tasks but unsafe for independent legal risk decisions.
- The article stresses that liability-bearing judgements must be based on expert-authored assessment rules and trusted data.
- Moverly's DiligenceEngine is described as using a playbook of more than 300 checks and 2,000 scenarios.
- The initial use case is auction buyers, where legal packs are often reviewed without full legal advice before bidding.
- The author says the real test is how AI systems are measured against expert judgement, not just how advanced the model is.
This is an RPSA summary of a publicly available article. The full content remains with the original publisher.
