StartupsNov 3, 2026, 7:04 PM UTC·2 min read

Trajectory raises $45M Series A led by Sequoia at $300M valuation

US-based AI platform Trajectory has closed a $45 million Series A led by Sequoia, with Nvidia also participating, valuing the company at $300 million.

Abstract illustration of an AI model learning loop with feedback arrows, representing Trajectory's continual learning platform. — Image: Webrazzi

SUMO

Dialo

ediviewer

Trajectory, a US-based AI platform that helps companies adapt models to their own work, has raised a $45 million Series A round led by Sequoia. Nvidia, Bessemer Venture Partners and Bain Capital Ventures also joined the round, which values the company at $300 million.

The startup says it will use the money to build out its platform, improve continual learning and post-training of AI models, and support its growth.

Retraining models on real feedback

Trajectory was founded by Ronak Malde, Michael Elabd and Arjun Karanam. Its pitch is that general-purpose models often fall short in specialized fields such as law and finance, and that companies should be able to make them better at specific tasks without building in-house research teams or paying for expensive engineering consultants.

The platform connects open-source AI models to the applications a company already uses and collects feedback on how those models perform. It then uses that feedback to retrain the models. This approach, known as continual learning, lets a model improve in a given domain over time by drawing on past tasks and user input.

Aimed at non-technical teams

Trajectory says people without technical expertise can manage the training process. In law, for example, practitioners can work to improve how a model reviews contracts, while finance specialists can target company valuation performance.

The platform is already used by companies including Harvey, Rogo, Decagon, Mercor and Nvidia. The goal, according to the company, is for AI models to handle specialized tasks more efficiently and with more accumulated experience.

For businesses, the appeal is cost and control: tuning open-source models on their own feedback data could narrow the gap between off-the-shelf AI and the demands of regulated, expertise-heavy work. The backing of Sequoia and Nvidia also signals investor appetite for tools that focus on post-training rather than building new foundation models from scratch.