June 28, 2023

 Jason Block

 Neetika Narula

CLM contracts Consulting elm chatgpt contracts insights large language models contracts lifecycle management

As evidenced by a recent article in The Wall Street Journal, the surge in news coverage of advanced large language models (LLM), like that of OpenAI’s ChatGPT, has increased attention on the importance of data management. This is because training LLM-based tools requires ingesting vast amounts of structured data. In the case of contracts analysis (among the first law-related use cases for LLM), that training begins with organising and converting largely unstructured contracts into a structured and digestible format.

Long before the flurry of articles about ChatGPT, companies were already using Elevate’s expert-led, software-powered Contracts Insights service to systematically transform unstructured contract data and documents into formats suitable for LLMs. To do so, Contracts Insights combines proprietary artificial intelligence with human expertise in the loop, the latter being crucial to ensuring the highest quality results.

While human review is critical to delivering high-quality structured data to our customers (we leverage lawyers with at least five years of experience reviewing complex contracts), Elevate’s Analyse Documents software forms the foundation of our Contracts Insights service offering. Indeed, each step of the process leverages Analyse Documents, starting with ‘smart’ work assignments (whereby it routes a particular family of contracts for review to lawyers with the appropriate combination of experience, skill, and training), onto our multi-layered contracts review workflow, then to Quality Control, and ending with automated batch Quality Assurance.

As part of our workflow, Analyse Documents conducts first-pass automated extraction of metadata using our proprietary AI that incorporates algorithms trained through processing over 12 million contract documents and hundreds of millions of metadata fields. Analyse Documents then applies automated rule-based data validations defined by our team of solution experts (for example, if the ‘term’ of a contract is ‘evergreen’, the tool automatically detects an error if a reviewer tries to enter a date in the ‘expiration date’ field). Lastly, the software detects anomalies in data consistency once our expert lawyers have performed their quality control review as a final quality assurance step (e.g., the software will highlight instances where the customer’s organisation appears to be the counterparty or where the effective date of an amendment precedes that of the original contract).

Applying these battle-tested processes and software developed over a decade of providing this service, Elevate’s Contracts Insights provides our customers with an extremely high-quality, interrogatable set of structured contracts data. This enables our customers to learn from and appropriately react to the positions they have taken in their contractual negotiations, arming them with data for use in post-merger integration settings, regulatory compliance, and potential migration to Contracts Lifecycle Management systems. It’s little surprise then that an array of organisations seeking to create contracts LLMs have sought out Elevate and Contracts Insights to assist with training their contracts models. For many, the potential of LLM and AI for contracts is a new world – for Elevate, transforming contracts source material into useful, structured output is something we’ve done for years and continue to provide to our customers every day.