Trajectory Raises $45 Million Series A to Automate AI Model Training
Updated
Updated · TechCrunch · Nov 2
Trajectory Raises $45 Million Series A to Automate AI Model Training
1 articles · Updated · TechCrunch · Nov 2
Summary
$45 million in new funding values Trajectory at $300 million, giving the year-old startup capital to expand a platform that retrains AI models from user feedback.
The company targets a costly bottleneck in specialized AI: businesses often need in-house researchers or forward-deployed engineers to fine-tune models for fields like law and finance.
Trajectory says its continual-learning system connects to customer applications, collects performance feedback, and post-trains open-source models so lawyers, doctors, and analysts can improve them with less technical work.
Harvey, Rogo, Decagon, Mercor and Nvidia are already customers, and Trajectory is now opening the post-training platform to the public.
Sequoia led the round with Nvidia, Bessemer and Bain Capital Ventures participating, as startups race to automate post-training against rivals including Thinking Machines Lab and Applied Compute.