Trajectory Secures $45M Series A to Automate AI Model Fine-Tuning for Non-Tech Users

Trajectory announced a $45 million Series A financing that values the company at $300 million. The round was led by Sequoia Capital and included participation from Nvidia, Bessemer Venture Partners and Bain Capital Ventures. Founded roughly a year ago by three Stanford classmates with prior stints at Google DeepMind and Apple, the 18-person team aims to simplify the process of adapting large language models for niche professional tasks.

Current foundation models, whether from OpenAI, Anthropic or open-source projects, lack the ability to learn from their own errors without additional training. Enterprises that need domain-specific performance typically resort to fine-tuning the models, a step that often requires dedicated research teams or costly forward-deployed engineers. Trajectory’s founders argue that this barrier makes specialized AI adoption prohibitively expensive for many organizations.

The startup’s product implements a method known as continual learning. Clients begin with an open-source model, integrate Trajectory’s platform into their existing applications, and allow the system to capture feedback on model outputs. That feedback is then used to automatically retrain the model, gradually improving its accuracy on tasks such as contract review or financial valuation. Early adopters, including legal-tech firms Harvey and Rogo, as well as Decagon, Mercor and Nvidia, report that the approach lets subject-matter experts shape model behavior without writing code.

Trajectory is not alone in pursuing automated post-training solutions. Competitors such as Thinking Machines Lab, the team behind the Tinker fine-tuning platform, Applied Compute and Prime Intellect also target the same market. The company differentiates itself through its design-first interface, which co-founder Arjun Karanam says makes the workflow accessible to non-technical users. The team’s deep research pedigree, spanning DeepMind, Apple’s generative model work, OpenAI, Stripe and Netflix, provides credibility in both algorithmic development and product usability.

Looking ahead, Trajectory’s funding positions it to expand its platform and broaden its customer base across regulated industries. For founders and operators, the availability of a low-code continual learning tool could reduce reliance on specialist engineers and accelerate time-to-value for AI projects. The market will watch whether the company can maintain safety safeguards while scaling autonomous model improvement, a question that will shape investor confidence and industry adoption in the coming years.