Hundreds of South Korean law firms are now using a legal AI that reads entire case files - thousands of pages of indictments, financial records, and chat logs - and flags which witnesses are lying and where statements shift. AiLex, built by legal tech startup A2D2, analyzes the evidentiary record itself rather than just searching precedents. As of July, more than 400 law firms and 800 lawyers had subscribed, with 638 cases processed since its January launch.
A2D2 CEO Jang Il-jun told ChosunBiz that unlike conventional legal AI tools, AiLex reads the case. The system ingests PDFs of indictments, witness statements, financial transaction histories, and KakaoTalk messages. It then digitizes the materials, builds a case summary, and maps every person, statement, and piece of evidence.
"We incorporated the practical experience of senior lawyers who have handled numerous civil and criminal cases into the development process," Jang said.
How the AI caught statement changes in "The Glory"
To demonstrate the system, A2D2 fed AiLex the full case record from the Netflix drama "The Glory," a story about school violence and revenge. The AI read thousands of pages of fictional case records and cross-checked character statements against evidence. When Park Yeon-jin first said "I have never assaulted anyone," the AI tracked the change as accomplices confessed. Park then said it was "only a scuffle." When medical records of a burn injury surfaced, she claimed: "A hair straightener did touch the body, but there was no intent to injure." AiLex assessed her credibility as low. By contrast, it rated Moon Dong-eun's statements as consistent with medical records and physical evidence.
What makes AiLex different from precedent-search AI
The core distinction is document analysis rather than statutory research. "Other legal AIs have focused on searching precedents and statutes," Jang said. "AiLex specializes in analyzing the evidentiary record and facts of individual cases."
The system classifies each uploaded document by type - indictment, evidence, third-party statement, suspect statement - then ranks content by importance. Lawyers still make final judgments. "The final judgment and defense strategy decisions rest with the lawyer," he said.
How processing thousands of pages works
AiLex relies on proprietary memory management technology. Instead of having the AI memorize massive documents as raw text, the system digitizes documents page by page, identifies their interrelationships, then splits facts into small units and stores them by topic. It uses knowledge graph technology to structure relationships among data, reducing memory load.
Jang said dozens of lawyers reported that record review time has dropped significantly. In large criminal cases with same suspects questioned multiple times, the volume can reach tens of thousands of pages. "By using AiLex, you can quickly identify important materials and issues without reading the entire record from start to finish," he said.
Agent AI for defense strategy
A2D2 has added agent capabilities to AiLex. The system can answer questions like: "Who might be a witness who would testify favorably for me?" It reanalyzes the entire record to find testimony supporting the defense's argument. It can also locate statements about coercion or threats in interrogation records, or present evidence that another defendant played a leading role.
Asked whether legal AI might replace lawyers, Jang said the opposite. "Rather than replacing lawyers, the goal is to expand the range of cases lawyers can take." Many cases have been rejected because fees were low relative to time required. "But if reduces the reduce time for record review and paperwork, lawyers can take on more cases at lower cost. Consumers can access legal services more cheaply, and lawyers can approach new markets."
On data security, AiLex operates in an isolated environment. User-uploaded records are not used for training, and the development team cannot view client records. Training uses separate statutory materials and public precedent databases.
Why this matters for legal professionals
For lawyers and supporting staff who spend days or weeks reviewing discovery and case filings, window into how quickly large-text analysis is evolving. The technology that reduces a 10,000-page file to a structured summary marks a practical shift in workflow. AI Learning Path for Paralegals and other resources can help legal teams understand these tools and integrate them into practice. Anyone who manages document-intensive litigation will likely face pressure to adopt similar technology as clients demand faster, lower-cost case work.
Your membership also unlocks:
