Jacob Lauritzen
Chief Technology Officer (CTO) of Legora, responsible for the technical architecture and engineering practices of the company's legal AI products.
Timeline (last 90 days)
Legora CTO Jacob Lauritzen said that Legora cannot read any trace of what the agent does for lawyers, because the data belongs to the law firm's clients.
Legal AI faces the "evals without traces" problem: lawyer workflows are highly confidential, interaction data is hard to retain, and users rarely proactively label errors, resulting in a lack of trace data for evaluation.
In legal AI, evaluation should focus on task-level correctness (such as whether clause extraction misses anything, whether cited statutes actually exist, and whether conclusions are consistent with the documents) rather than conversation-level satisfaction.
Lauritzen observed that AI's erosion of the billing model is not simply "AI replacing lawyers" but a misalignment in how value is measured: AI compresses ten hours of work into one hour, and if law firms continue to bill by the hour their revenue structure will loosen, while shifting to outcome-based or value-based billing requires rebuilding pricing and trust mechanisms.
Advice for vertical AI entrepreneurs includes: design the evaluation system around task correctness from day one, reserve capabilities in the system architecture for batch and structured tasks, and understand that understanding the customer's business model defines product direction more than understanding the workflow.
All Facts (5)
Legora CTO Jacob Lauritzen said that Legora cannot read any trace of what the agent does for lawyers, because the data belongs to the law firm's clients.
50%UnverifiedLegal AI faces the "evals without traces" problem: lawyer workflows are highly confidential, interaction data is hard to retain, and users rarely proactively label errors, resulting in a lack of trace data for evaluation.
50%UnverifiedIn legal AI, evaluation should focus on task-level correctness (such as whether clause extraction misses anything, whether cited statutes actually exist, and whether conclusions are consistent with the documents) rather than conversation-level satisfaction.
50%UnverifiedLauritzen observed that AI's erosion of the billing model is not simply "AI replacing lawyers" but a misalignment in how value is measured: AI compresses ten hours of work into one hour, and if law firms continue to bill by the hour their revenue structure will loosen, while shifting to outcome-based or value-based billing requires rebuilding pricing and trust mechanisms.
50%UnverifiedAdvice for vertical AI entrepreneurs includes: design the evaluation system around task correctness from day one, reserve capabilities in the system architecture for batch and structured tasks, and understand that understanding the customer's business model defines product direction more than understanding the workflow.
50%