The French artificial intelligence company Mistral AI has launched Forge, a platform designed to help organizations develop AI models using their own proprietary data rather than relying solely on public datasets. The system integrates an enterprise’s internal documentation, engineering standards, and operational processes to ensure that AI agents understand specific institutional contexts and terminologies.
Designed with an agent-first approach, Forge enables autonomous systems to manage model fine-tuning and optimization through natural language. Rather than a static deployment, the system is built for continuous updates to reflect changing regulations and new data. This will allow sectors like finance, government, and software engineering to maintain AI models as evolving assets that reflect current institutional knowledge and operational requirements, as stated in their announcement.
By allowing models to be trained and operated within an organization’s own infrastructure, Forge addresses concerns regarding data control and intellectual property, ensuring that AI outputs remain aligned with internal compliance and governance standards.
Forge has already been adopted by several high-profile organizations, including ASML, Ericsson, and the European Space Agency.
