Trajectory, an AI startup founded by three Stanford classmates, has raised $45 million in a Series A funding round. The deal values the 18-person company at $300 million.
Sequoia led the round. Nvidia, Bessemer Venture Partners, and Bain Capital Ventures also took part.
The company was founded about a year ago. Its founders have worked at Google DeepMind and Apple.
Trajectory builds tools that help AI models keep learning after they are put to use. The goal is to let experts without technical training improve models for their own work.
Why AI Models Need More Training
Current AI models, whether built by OpenAI, Anthropic, or on open source alternatives, share a common limit. They do not learn from their mistakes without further training.
Businesses that use AI in fields like law and finance often fix this by fine-tuning general models. That work usually requires hiring internal research teams or costly consultants known as forward-deployed engineers.
Co-founder Ronak Malde compared foundational models to a very smart PhD student who is always starting a new job. “But what the economy values is someone who worked on the job for 30 years,” he told TechCrunch.
Trajectory aims to solve this through an approach called continual learning. Businesses start with an open source AI model and connect the platform to their applications.
The platform collects feedback on how the model performs. Trajectory then uses that information to retrain the model so it improves at specialized tasks over time.
Malde said lawyers, doctors, and other specialists can improve the models almost automatically on their own. Users decide “what good looks like for their organization and can turn that into improvements for their model,” he said.
Customers, Rivals, and Team
Harvey and Rogo already use the platform to post-train open source models. The models learn from past tasks, such as reviewing a certain type of legal contract or valuing a company with a set method.
Other customers include Decagon, Mercor, and Nvidia. The aim is to help models perform more like experienced lawyers or financial analysts.
Trajectory’s method shares some traits with recursive self-improvement, the idea that AI systems can improve their own abilities. Co-founder Michael Elabd said keeping humans involved in training makes the approach safer, since fully autonomous self-improvement could lead to unintended or dangerous results.
The company faces competition from other startups working on post-training. These include Thinking Machines Lab, founded by former OpenAI CTO Mira Murati and maker of the fine-tuning platform Tinker.
Applied Compute, valued at $3.25 billion in October, is also in the market. Prime Intellect, which helps companies build and train their own AI agents, is another rival.
Malde said the startup’s biggest advantage is its team. He was among a handful of researchers Google DeepMind acqui-hired from Windsurf, and Elabd also worked at DeepMind.
Third co-founder Arjun Karanam previously worked on generative foundation models at Apple. The startup has also hired researchers and product designers from OpenAI, Stripe, and Netflix.
Karanam said the company has built an interface that makes post-training easier for nontechnical customers. Trajectory launched on November 2 and is now making its post-training platform publicly available to businesses.
