GC AI customers say its legal software saves teams 14 hours a week

In-house legal teams using GC AI reported saving an average of 14 hours a week, according to a customer survey disclosed in an Anthropic case study. The findings cover more than 100 active customers using software that combines Claude with company-specific legal standards and source-linked answers, while leaving lawyers responsible for reviewing the work.

The same survey found a reported 14% reduction in spending on outside counsel and 21% greater accuracy on legal tasks compared with general-purpose AI tools without legal workflow structuring. Those comparisons concern GC AI’s full product, rather than Claude used on its own.

The figures are customer-survey findings published by GC AI’s model supplier, not independently verified performance measurements. The case study does not explain how the accuracy improvement was scored or name the general-purpose tools used for comparison.

Anthropic says more than 1,500 companies use the platform, including Zscaler, News Corp and Nextdoor. GC AI handles legal research, contract review, document drafting and internal legal guidance. It routes questions and documents through task-specific workflows that incorporate a department’s approved positions, risk tolerance, preferred language and internal guidance.

We needed models that could produce clear, structured outputs that lawyers could review quickly and integrate into their work.

Cecilia Ziniti, GC AI co-founder and CEO

The product gives lawyers a route back to the material behind an answer. Its Exact Quote feature links conclusions directly to source text. Integrated search supplies legal materials and citations to statutes, regulations and official guidance. Lawyers check the reasoning and relevant sources before finalizing advice or documents.

GC AI also built a proprietary knowledge-management system to supply relevant documents, sections and metadata during each workflow. Ziniti says many users work with dozens of documents exceeding 100,000 tokens each, quickly filling the amount of text a model can handle at once.

GC AI uses Claude Sonnet for its main customer-facing workflows, citing its balance of performance, speed and cost. It reserves Opus for more complex reasoning. The company evaluates models on real legal tasks, with a key measure being how quickly a lawyer can turn an output into a final answer.

The software reaches lawyers through a web interface, a Microsoft Word plug-in and Google Drive connectivity. Company context, playbooks and prior work carry across those surfaces.