Trajectory has introduced a self-serve tool that enables users to customize and deploy artificial intelligence models based on their own specific data and quality standards. The company secured $60M in funding led by Sequoia to support the debut of the platform, which allows professionals such as engineers and doctors to post-train models of up to a trillion parameters, valuing the startup at $300M.
The launch includes research previews of GS1, an algorithm that allows AI models to learn directly from production activity, and a tool called Continual Grading designed to scale how a model identifies high-quality responses. The platform is intended to shift the control of model behavior and reward definitions away from a small number of AI labs to the individual users of the technology.
