A new AI tool has hit the market that allows professionals such as doctors and lawyers to customize the behavior of large language models without needing a background in computer science. The startup behind the tool, Trajectory, brought its total capital to $60M following a recent $45M Series A round led by Sequoia that valued the company at $300M. The lauch focuses on a self-serve platform that handles the training, evaluation, and deployment of trillion parameter models using a user’s own production data and as few as five lines of code.
As part of the release, Trajectory introduced two research previews: a tool called Continual Grading to help scale human preference standards, and a new algorithm, GS1, that allows models to learn directly from their activity in production. The company says these tools are intended to shift the control of model behavior and quality definitions away from a handful of AI labs and into the hands of the individual users.
