Legora Reviews 41 Documents in Minutes with GPT-6 Astra, Boosting Accuracy by 40%

The agentic operating system for legal and professional services, Legora, has integrated GPT-6 Astra to significantly enhance the performance of complex financial statement tie-out workflows. Legora serves over 1,800 legal departments and law firms across more than 50 global markets, and this update reduces tasks that previously took hours or days to just minutes.

Large-Scale Document Processing and Improved Accuracy

Legora's agents review 41 documents in a single run, cross-referencing each item against trial balances, consolidation schedules, and prior-year accounts to identify and document discrepancies. In Legora's proprietary Agentic Reasoning (BAR) benchmark evaluation, GPT-6 Astra demonstrated a 40% performance improvement in financial statement workflows compared to previous models. The average improvement across all tasks was approximately 3%.

Notably, the agent identified all 4 errors that Legora had intentionally planted, including a £500,000 discrepancy hidden in the revenue notes. The agent cross-references all balances with supporting schedules and logs each check, ensuring that final judgments remain with legal professionals.