Startup Trajectory raises $45 million for a platform for automated post-training of AI models
Startup Trajectory raised $45 million in a Series A round led by Sequoia at a $300 million valuation and launched a platform that automates the post-training of AI models from feedback, without requiring programming.
Trajectory, a startup founded a year ago by three Stanford classmates with prior experience at Google DeepMind and Apple, has announced closing a Series A funding round of $45 million at a valuation of $300 million. The round was led by Sequoia Capital, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures. The company has 18 employees.
Trajectory's platform allows companies to connect an open source AI model to their own applications, collect feedback on its performance, and based on that automatically fine-tune (post-train) the model so that it gradually improves on specialized tasks. According to the company, this allows specialists, such as lawyers or financial analysts, to determine for themselves what counts as a "good" outcome and translate that into model improvements without involving an in-house research team or external consultants (forward-deployed engineers).
According to the company, the platform's customers include Harvey, Rogo, Decagon, Mercor, and Nvidia, who use it to fine-tune open source models for tasks such as reviewing specific types of contracts or valuing companies using a specific methodology. The platform is now publicly available. Trajectory considers its direct competitors to be Thinking Machines Lab (the Tinker platform, founded by Mira Murati), Applied Compute (according to the source, valued at $3.25 billion last month), and Prime Intellect.
Co-founder Michael Elabd says he distinguishes the company's approach from the concept of recursive self-improvement of AI by the fact that a human remains involved in the training process, which he says makes the approach safer.
Why it matters
For companies deploying AI in fields such as law or finance, this represents an alternative to building an in-house research team or purchasing forward-deployed engineer services – according to Trajectory, it allows specialists without programming knowledge to gradually improve a model for their own specific tasks. The size of the funding and the involvement of investors such as Nvidia or Sequoia also points to growing interest in the market for post-training and continual learning tools, where Trajectory competes with Thinking Machines Lab, Applied Compute, and Prime Intellect.
Relevant practical impact
What this means
For a business
Companies that want to deploy specialized AI models (e.g. in law or finance) can, according to the source, use the Trajectory platform to fine-tune open source models based on feedback from their own specialists, without needing an in-house research team or external consultants.
DevelopmentCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.