Pattern Recognition vs Structured Extraction

This split shows that legacy contract AI won adoption through two very different buyer instincts, discovery versus control. Luminance was built to ingest a pile of contracts and automatically cluster unusual language, which fit teams hunting for hidden risk in messy diligence. Kira won firms that wanted the software to pull the same known fields, like assignment, change of control, or termination, from every document in a repeatable way, then feed that output into deal checklists and review workflows.

• Kira became the structured option because it was designed around lawyer trained extraction models. Litera describes Kira as delivering contract review with 1,400 plus proprietary fields and 90% plus extraction accuracy, which maps to firms that want a defined list of clauses surfaced every time, not an open ended anomaly scan.
  3 businesswire 4 litera
• Luminance built its pitch around low setup pattern recognition. Company materials describe unsupervised machine learning that reads a document set without prior training, and broader research shows the product later expanded from diligence into drafting, redlining, and negotiation through Word based workflows and enterprise subscriptions.
  1 sacra 5 luminance 6 luminance
• The market rewarded both approaches, but for different jobs. Baker McKenzie chose eBrevia over both vendors in 2017 after pilots, showing buyers often picked the tool whose workflow matched the contract type and review process best, not a single universal winner across all legal work.
  1 sacra 7 legaltechnology

Going forward, the old anomaly versus extraction divide matters less as contract AI turns into full workflow software. Kira is being folded deeper into Litera’s broader transaction stack, while Luminance is extending from review into negotiation and execution. The advantage will come from owning the whole path from document intake to final action, not just spotting clauses faster.