This move marks a shift for the company, which previously built its $11 billion business exclusively by utilizing third-party AI models.
Developing an in-house model allows Harvey to reduce its dependence on external providers like OpenAI and Anthropic, who have recently begun competing for legal industry clients.
By routing tasks through Tenet instead of paying fees for every request made to outside models, the company aims to lower its operational costs and improve profit margins.
Harvey co-founder Gabe Pereyra noted that the bespoke model is intended to handle complex legal reasoning and tasks that typically take lawyers hours or days to complete, providing a specialized alternative to general-purpose AI.
While Tenet is not yet live for general users, it is part of a broader "Harvey II" update that includes a "Memory" feature for saving specific user preferences across different tasks.
Moving forward, Harvey intends for Tenet to serve as a foundational layer that law firms can use to train their own custom models based on their internal case history and institutional knowledge.
Although the company plans to release research benchmarking Tenet’s performance against competitors, it has not yet disclosed which law firms are testing the technology or a specific date for its full release.