Trajectory, a startup founded about a year ago by three Stanford classmates with experience at Google DeepMind and Apple, has raised $45 million in a Series A round at a $300 million valuation. Sequoia led the round, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures. The 18-person company is building a platform that automates post-training for AI models, so businesses can adapt open source models to specialized fields like law and finance without hiring internal research teams or relying on expensive forward-deployed engineers.

The core problem Trajectory addresses is that current AI models do not learn from their mistakes without further training. Businesses typically fine-tune general models for specialized tasks, but that process is costly and requires expertise. Trajectory applies an approach called continual learning: customers connect the platform to their applications, starting with an open source model, and the platform collects feedback on how the model performs. Trajectory then uses that feedback to retrain the model, improving its performance on specialized tasks over time. Co-founder Ronak Malde compared foundational models to a very smart PhD student who is always starting a new job, while what the economy values is someone who has worked on the job for 30 years.

The company says lawyers, doctors, and other specialists can use Trajectory to improve models nearly automatically, potentially making specialized AI deployment faster and cheaper. Existing customers include Harvey, Rogo, Decagon, Mercor, and Nvidia. Harvey and Rogo use the platform to post-train open source models so they perform more like experienced lawyers or financial analysts by learning from previous tasks, such as reviewing a particular type of legal contract or valuing a company using a specific methodology. Trajectory is now making its post-training platform publicly available. According to Malde, the people using Trajectory decide what good looks like for their organization and can turn that into improvements for their model.

Trajectory’s continual learning approach shares some similarities with recursive self-improvement (RSI), where AI systems improve their own capabilities. However, co-founder Michael Elabd argues that keeping humans involved in the training process makes Trajectory’s approach safer, since autonomous self-improvement could lead to unintended or dangerous outcomes.

The startup faces competition in the automated post-training space. It considers Thinking Machines Lab — founded by former OpenAI CTO Mira Murati and the company behind the fine-tuning platform Tinker — a direct competitor, along with Applied Compute, valued at $3.25 billion last month, and Prime Intellect, which helps companies build and train their own AI agents.

Malde believes Trajectory’s biggest advantage is its team. He was among the handful of researchers Google DeepMind acqui-hired from Windsurf. Elabd also worked at DeepMind, while the third co-founder, Arjun Karanam, previously worked on generative foundation models at Apple. The startup has also recruited researchers and product designers from OpenAI, Stripe, and Netflix. Karanam says the company developed an interface that makes post-training more accessible to its nontechnical customers, which is why design matters to an AI training company.