Case Study
Prioritising Document Review and Increasing Speed to Insight with AI
An Australian law firm was tasked with quickly classifying thousands of documents into relevant categories for a client’s dispute. FTI Technology was engaged to support the review and apply IQ.AI by FTI Technology to improve speed and precision with the use of artificial intelligence.
Situation
With only two weeks until the trial date, the client received a set of more than 10,000 documents from opposing counsel that required classification into various categories of complaints. Upon review of an initial batch of documents, the legal team realised that many were not relevant to the matter. They needed a way to efficiently prioritise the documents by specific severity and relevance parameters, so that the legal team could finalise case strategy and fully understand the content of the materials prior to the impending trial.
Our Role
FTI Technology’s experts collaborated with the legal team to build a large language model prompt within IQ.AI for Review, to classify documents into four specific issues. Each specific issue would also have a rating extracted based on its severity. Additionally, the AI workflow was designed to provide details including the reasoning for why a document was classified to a certain category and an excerpt directly taken from the document to validate the reasoning.
A small sample of documents were reviewed using IQ.AI and the results were compared to the results of manual review on the same set of documents. IQ.AI demonstrated a 96% accuracy rate for classifying documents to the correct issues and severity. For the 4% that were not accurate, the reasoning and excerpts provided clarification as to why the AI considered the document to be relevant.
FTI Technology’s experts further refined and applied similar prompts across the entire document set, enabling highly accurate prioritisation of the 900 documents that were most severe and relevant to the case.
In some instances, IQ.AI tagged documents as high priority, while noting in the reasoning explanation that the document may not be within scope but was still of interest given mentions of concepts that were considered relevant. The legal team found that while some of these documents were not technically relevant based on severity, they were still of interest to provide additional understanding of the matter.
Our Impact
FTI Technology optimised document review for the client using IQ.AI, providing a 96% accuracy rate for document categorisation. The insights from this accelerated review supported the legal team in validating its position in the dispute, which led to opposing counsel dropping the case.
Expert-led, AI-enabled workflows increased speed to insight and drastically reduced the amount of time needed for review, allowing the legal team to prepare for trial under a compressed timeline and ultimately succeed in its pursuit of the case’s dismissal.