Rebuilding Harvey’s review algorithm to increase accuracy and speed
Review tables now deliver more accurate answers, more granular citations, and faster results.
Review tables in Harvey enable legal teams to extract information from large sets of documents at once — turning thousands of contracts, filings, or agreements into a structured grid you can scan, filter, and act on. For tasks like due diligence, contract analysis, or regulatory review, it compresses work that would otherwise take days into something you can move through in minutes.
Earlier this year, we revamped the algorithm underlying review tables so that they deliver more accurate answers, more granular citations, and faster results. While our previous review algorithm showcased the power of extraction en masse, the new algorithm extends the value of review tables with a focus on empowering users to efficiently review at scale.
In this post, we’ll walk through the problems we set out to solve, what we changed, and how we evaluated the impact of those changes.
To read the article in full, click here.
