Trajectory is an AI infrastructure company that provides a continual learning platform enabling AI-native B2B software teams to capture real user feedback signals—such as corrections, edits, and retries—from production and automatically use them to improve their AI models, prompts, and agent harnesses on an ongoing basis. Teams integrate via a lightweight SDK that standardizes agent traces and product telemetry into a training primitive, with customers including Harvey, Clay, Decagon, and Mercor.
“Trajectory is building a continual-learning platform so AI-native companies can keep models improving after deployment instead of retraining from scratch... it bets on post-training learning as core infrastructure, not a research side project.”
Source→“Trajectory raised a $40M Series A, Sequoia-led with Bessemer and Radical Ventures. Angels include Jeff Dean and Fei-Fei Li.”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.