Grok Leads Contract Review Benchmark

Ivo and data lab micro1 tested seven AI models on 350 real-world contract review tasks to compare their decisions with those of experienced attorneys. The benchmark assessed whether models prioritized important issues, made appropriate edits and recognized when a decision needed human input. All seven scored below 60% on weighted attorney-defined criteria; the highest score was 58.3%.

Grok 4.7 ranked first on weighted rubric pass rate at 58.3%, followed closely by Opus 5.5 at 58.1% and Kimi K3 at 57.1%, according to Ivo. Completion rates ranged from 80.8% for Opus to 99.5% for Grok and 100% for Kimi. Kimi cost 38% less per completed rollout and finished 13% faster than Grok. The article says complex legal work also needs matter-specific reference data, such as rulebooks and precedent documents, plus oversight that can correct initial outputs.

Amsterdam-based legal AI platform Saga launched Saga External, a tool for teams to create legal services from their own playbooks and make them available to clients. Examples include a contract checker based on a firm’s playbook and a compliance review. Ertan Bera, who worked on core AI product development at Lovable, joined Sweden’s Aloi as Applied AI Lead; Erik Lindberg also joined as CRO after five years as Sana’s VP of Sales through its billion-dollar exit. Aloi is developing a judgment layer for law firms using past matters and work performed on them.

Checkbox launched Matter History, a record of changes to legal matters and who or what made them, including lawyers and AI agents. Other features include deadline tracking, bulk editing across matters and multi-document DocuSign envelopes. Lawfecta said it is developing an AI-powered operating system for attorneys to organize workflows, prepare tasks for delegation and connect with professionals who can do the work. Mary Technology announced an integration with Harvey: Mary maps uploaded matter documents before the first question and retains findings and review as a case develops, with that accumulated understanding made available in Harvey. Littler announced partnerships with Legora and Epiq to help employers navigate legal, workforce and operational implications of AI-enabled workforce transformation. BLEN said its Federal Register and eCFR connector for Google Cloud’s Gemini for government services is available, giving agents read-only access to official US federal regulatory data from all three sources.

Artificial Lawyer said it would pause news publishing until Thursday, October 29, while its founder was in Boston for ClioCon’s Enterprise track, then attend Legal Innovators UK on November 4 and 5 and Legal Innovators New York on November 17 and 18. The site also advertised a Harvey webinar on legal agents for November 2 at 11 am ET and 4 pm GMT; the 45-minute session was to cover use cases, human review and rollout. A Spellbook webinar and live demonstration of its autonomous contract management capability was scheduled for Thursday, October 29 at 12:00 PM ET and 5PM UK.

Ivo and data lab micro1 tested seven AI models on 350 real-world contract review tasks to compare their decisions with those of experienced attorneys. The benchmark assessed whether models prioritized important issues, made appropriate edits and recognized when a decision needed human input. All seven scored below 60% on weighted attorney-defined criteria; the highest score was 58.3%.

Grok 4.7 ranked first on weighted rubric pass rate at 58.3%, followed closely by Opus 5.5 at 58.1% and Kimi K3 at 57.1%, according to Ivo. Completion rates ranged from 80.8% for Opus to 99.5% for Grok and 100% for Kimi. Kimi cost 38% less per completed rollout and finished 13% faster than Grok. The article says complex legal work also needs matter-specific reference data, such as rulebooks and precedent documents, plus oversight that can correct initial outputs.

Amsterdam-based legal AI platform Saga launched Saga External, a tool for teams to create legal services from their own playbooks and make them available to clients. Examples include a contract checker based on a firm’s playbook and a compliance review. Ertan Bera, who worked on core AI product development at Lovable, joined Sweden’s Aloi as Applied AI Lead; Erik Lindberg also joined as CRO after five years as Sana’s VP of Sales through its billion-dollar exit. Aloi is developing a judgment layer for law firms using past matters and work performed on them.

Checkbox launched Matter History, a record of changes to legal matters and who or what made them, including lawyers and AI agents. Other features include deadline tracking, bulk editing across matters and multi-document DocuSign envelopes. Lawfecta said it is developing an AI-powered operating system for attorneys to organize workflows, prepare tasks for delegation and connect with professionals who can do the work. Mary Technology announced an integration with Harvey: Mary maps uploaded matter documents before the first question and retains findings and review as a case develops, with that accumulated understanding made available in Harvey. Littler announced partnerships with Legora and Epiq to help employers navigate legal, workforce and operational implications of AI-enabled workforce transformation. BLEN said its Federal Register and eCFR connector for Google Cloud’s Gemini for government services is available, giving agents read-only access to official US federal regulatory data from all three sources.

Artificial Lawyer said it would pause news publishing until Thursday, October 29, while its founder was in Boston for ClioCon’s Enterprise track, then attend Legal Innovators UK on November 4 and 5 and Legal Innovators New York on November 17 and 18. The site also advertised a Harvey webinar on legal agents for November 2 at 11 am ET and 4 pm GMT; the 45-minute session was to cover use cases, human review and rollout. A Spellbook webinar and live demonstration of its autonomous contract management capability was scheduled for Thursday, October 29 at 12:00 PM ET and 5PM UK.