How does $45M Series A enable non-engineers to retrain AI via continual learning?

TechCrunch reported that Trajectory, a startup founded by three Stanford alumni with backgrounds at Google DeepMind and Apple, has raised $45 million in a Series A round.reported The funding round was led by Sequoia Capital, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures.Trajectory, a startup helping non-engineers train AI models, has raised $45 million in new funding.Trajectory secured $45M in Series A funding for its automated continual learning platform, targeting post-training of open-source models. This investment values the 18-person company at $300 million, marking a significant early milestone for the one-year-old venture focused on specialized AI model training.reported

The announcement

Trajectory aims to remove the engineering barrier from AI model fine-tuning. Co-founder Ronak Malde told TechCrunch that current foundational models resemble smart PhD students starting new jobs, lacking the accumulated experience of industry veterans. The platform allows specialists to connect their applications to open-source models, collecting performance feedback to automatically retrain the system. This continual learning approach helps models improve at specific tasks over time without requiring manual intervention from data scientists.

The context

Businesses adapting general AI models for fields like law and finance typically rely on expensive forward-deployed engineers or internal research teams to fine-tune models. Trajectory’s platform automates this process by using user feedback to guide retraining. Companies such as Harvey and Rogo are already using the tool to post-train open-source models, helping them perform tasks like contract review or company valuation more effectively. Other customers include Decagon, Mercor, and Nvidia, according to the report.reported

Reading between the lines

The startup’s approach shares similarities with recursive self-improvement, where AI systems enhance their own capabilities. By allowing users to define what good performance looks like for their organization, Trajectory turns subjective organizational standards into concrete model improvements. This shift suggests a move toward more autonomous model maintenance, reducing the dependency on specialized technical staff for routine model updates and performance optimization.

The business case

Trajectory’s platform is now publicly available, potentially making specialized AI deployment faster and cheaper. By enabling lawyers, doctors, and other professionals to improve models themselves, the startup addresses the high cost of hiring forward-deployed engineers. For AI platform architects, this offers a vendor option that reduces reliance on expensive consultants for model fine-tuning, though the long-term efficacy of automated continual learning in production environments remains to be fully validated by wider market adoption.