AI Training Startup Trajectory Secures $45 Million in New Funding

A one-year-old artificial intelligence startup called Trajectory has raised 45 million dollars in a Series A funding round, bringing its total valuation to 300 million dollars. Sequoia led the investment round, with additional backing from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures. Three former Stanford classmates who previously worked at Google DeepMind and Apple founded the 18-person company.

Platform And Purpose

Current AI models created by companies like OpenAI and Anthropic struggle to learn from their mistakes without extra training. Businesses adapting these models for specialized fields such as law and finance usually need to hire internal research teams or expensive consultants known as forward-deployed engineers. Trajectory automates this process through an approach called continual learning. Users connect the platform to their applications to collect performance feedback, which Trajectory then uses to retrain the model for specialized tasks.

Customers And Competition

Lawyers, doctors, and other specialists can use the platform to improve models almost automatically without needing technical coding skills. Existing customers using the software to post-train open-source models include Harvey, Rogo, Decagon, Mercor, and Nvidia. Co-founder Ronak Malde noted that the system lets users define quality standards for their own organizations. Competitors in the automated post-training space include Thinking Machines Lab, Applied Compute, and Prime Intellect.

Context

Continual learning helps bridge the gap between general-purpose artificial intelligence and the specific needs of professional industries. By keeping humans involved in the training loop, the platform aims to avoid the safety risks associated with fully autonomous self-improvement. The startup is now making its post-training platform available to the public.