Trajectory raises $45M to help experts train AI models without coding

In short: Trajectory raised $45 million to help companies improve AI models using everyday work feedback, without needing large technical teams.

What happened

Trajectory, a startup founded about a year ago, announced a $45 million Series A funding round. The round values the company at $300 million. It was led by Sequoia, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures.

The company’s goal is to make “post-training” easier. Post-training is the extra training companies do after they start with a general AI model, so it gets better at a specific job, like reviewing legal contracts or analyzing financial reports. Today, many companies do this by hiring specialized engineers or expensive consultants.

Trajectory says its platform lets businesses connect an AI model to their own apps and workflows. It collects feedback on how the model performs, then uses that feedback to retrain the model over time. Think of it like coaching a new employee by marking what they got wrong, then turning those notes into better performance next week.

TechCrunch reports that companies including Harvey and Rogo already use Trajectory to help open source models learn from past work, so the models behave more like experienced lawyers or financial analysts. Other customers mentioned include Decagon, Mercor, and Nvidia. Trajectory is also making its post-training platform publicly available.

Why it matters

More companies want AI that understands their exact rules and standards, not just generic answers. If tools like Trajectory work as promised, specialists like lawyers and doctors may be able to improve AI systems themselves, which could reduce costs and speed up how quickly tailored AI gets deployed.

Source: TechCrunch AI