Trajectory Raises $45 Million Series A at $300 Million Valuation

Trajectory has raised $45 million in Series A funding at a $300 million valuation, money the company will spend pushing post-training infrastructure into law firms, hospitals and engineering shops that have never built a machine learning system.

Sequoia Capital led the round, announced November 2, with participation from NVIDIA, Bessemer Venture Partners and Bain Capital Ventures. The financing brings Trajectory's total funding to $60 million and lands on the same day the company launched its AI model post-training platform.

The bet is straightforward: general-purpose models don't know what a good result looks like inside a specific profession. Trajectory's platform lets organizations train AI models around their own tasks, professional knowledge and performance requirements, instead of renting behavior from a generic system.

The platform handles the full cycle — training, evaluation and deployment — including models with up to a trillion parameters. Users kick off those workflows through an interface that requires only a small number of AI agent prompts, no in-house ML team required.

That last point is the commercial wedge. Trajectory is targeting lawyers, doctors, engineers and financial analysts — people who know exactly what a correct output looks like in their daily work but have no specialized experience training models. The company argues these professionals are best positioned to improve AI performance because they already hold the evaluation criteria.

Who is already using it?

The customer list gives the thesis early proof. Harvey, Rogo, Mercor and NVIDIA are using Trajectory's technology to train models for their own needs, according to the company. NVIDIA's presence is notable — it's both a customer and an investor.

For Trajectory, those relationships signal demand for enterprise AI infrastructure that gives organizations more control over model behavior and performance than off-the-shelf APIs allow.

What did the company launch alongside the funding?

Two research previews shipped with the platform, both aimed at continual learning — the idea that a model shouldn't stop improving after its initial training.

• Continual Grading scales a user's judgment. The system learns a user's preferences and quality standards from far more examples than any individual could manually evaluate, so the model absorbs what "good" means for a specific application.
• GS1 is a new algorithm that lets models learn directly from what they do in production environments. Instead of waiting for periodic retraining on manually prepared datasets, models improve through operational experience as it happens.

Founder Ronak Malde framed the mission in dollar terms: the round exists "to democratize the most powerful technology of today – training models – so that lawyers, doctors, and engineers of the world can own their own intelligence." His launch message to users was blunt: "You know what good looks like, now go build it. Own your own value function with Trajectory."

Why does model ownership matter now?

Trajectory's positioning rests on a simple claim: businesses should control not only the models they deploy but also the criteria used to evaluate and improve them. That distinction gets sharper as organizations push AI into specialized workflows where accuracy, domain expertise and consistent performance carry real risk.

The company plans to deploy the new capital across three fronts: advancing its post-training infrastructure, developing its continual learning technologies, and expanding adoption among enterprises building customized AI systems.

With $60 million raised and a $300 million valuation, Trajectory is wagering that the next phase of enterprise AI belongs to the people who own the expertise — not the ones who own the data centers.