Trajectory, a Stanford-founded startup, raised $45 million in Series A funding at a $300 million valuation to automate AI model fine-tuning for specialized industries. Sequoia led the round, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures.
The 18-person company, founded about a year ago by Ronak Malde, Michael Elabd, and Arjun Karanam, uses continual learning to let businesses collect feedback on AI model performance and automatically retrain models for tasks in law, finance, and other fields. This approach eliminates the need to hire research teams or expensive forward-deployed engineers.
"With Trajectory, lawyers, doctors, and other specialists can improve the models near automatically themselves," Malde said. The startup's interface is designed to make post-training accessible to nontechnical users.
Harvey, Rogo, Decagon, Mercor, and Nvidia already use Trajectory's platform. The startup is now making its post-training platform publicly available.
Trajectory's approach involves humans in the training loop, which co-founder Elabd argues is safer than autonomous self-improvement. The founders came from Google DeepMind, Apple, and other top tech companies.
Other startups pursuing similar post-training automation include Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, Applied Compute, valued at $3.25 billion last month, and Prime Intellect.