Harvey releases specialized model built for legal tasks

Legal technology startup Harvey announced a new artificial intelligence system called Tenet, designed specifically for complex legal work. According to a post from the company on X, Tenet builds upon an open source base model created by Moonshot AI and was fine tuned using legal documents, synthetic data, and expert feedback. Harvey also created specialized smaller tools within the system to handle specific tasks like corporate acquisitions and document review.

Specialized tools like Tenet show how artificial intelligence is shifting from general chatbots to highly tailored software for specialized industries. For everyday consumers and businesses, more efficient legal software could eventually lower legal fees and speed up lengthy processes like contract reviews or corporate due diligence. Furthermore, because Harvey claims the model operates at less than a quarter of the expense of general leading models, smaller law firms might soon gain access to advanced automation that was previously unaffordable.

It remains to be seen whether heavily customized models trained on synthetic legal data will perform reliably without introducing subtle hallucinations in high stakes court proceedings. As legal tech vendors move toward specialized subagents, the industry will need clearer standards to verify whether these automated systems meet professional liability requirements.