Spellbook splits drafting and review

This split shows that Spellbook is not just adding features, it is carving legal AI into two distinct jobs with different product surfaces. The Word add in is built for the moment a lawyer is inside one draft and wants redlines, comments, clause suggestions, and market checks without leaving Word. Associate moves up a level, where the work is comparing versions, updating a template from several precedents, or answering diligence questions across a folder of contracts in a separate workspace.1 2 3 4 5

• Associate is built around a file library and repeatable workflows. It can read Word files, PDFs, text files, spreadsheets, and compare or question them together, while only Word docs are directly editable. That is a different job from inline drafting inside a single contract.2 3 4
• The tables and repository layer turns contract sets into rows and columns, which is the shape lawyers need for diligence. Instead of reading 200 leases one by one, a team can extract fields like assignment rights or renewal terms into a grid and review exceptions in bulk.3 6
• Competitors are converging on the same divide. Harvey has a Word surface for drafting plus broader document system integrations, while Draftwise centers precedent search and insertion inside Word from firm repositories like iManage and NetDocuments. The common pattern is single doc work in Word, corpus level work in a separate retrieval layer.7 8 9 10

The next step is matter systems that feel less like chat and more like legal operating software. As Associate adds more connectors, saved workflows, and structured extraction, the winning products will own both the drafting moment in Word and the multi document review layer that sits over the firm’s whole contract corpus.2 3 5 7