Trajectory has raised $45 million in a Series A funding round led by Sequoia, giving the roughly year-old artificial intelligence startup a $300 million valuation as it develops tools for businesses that want to improve AI models without building large in-house research teams.
Nvidia, Bessemer Venture Partners and Bain Capital Ventures also participated in the round. Trajectory was founded by three Stanford graduates with experience at Google DeepMind and Apple. The company is targeting professionals and organizations that need AI systems to perform specialized work but may not have the technical staff required to retrain or fine-tune those systems themselves.
A feedback-driven approach to model improvement
Trajectory’s platform is built around a method it calls continual learning. The system connects open-source AI models to business applications, gathers information about how those models perform and uses that feedback to retrain them for particular tasks. The stated aim is to let organizations improve model performance over time rather than relying on a fixed system after deployment.
Trajectory’s approach addresses a limitation the company identifies in many existing AI models, including systems developed by OpenAI and Anthropic: they do not automatically learn from mistakes once they are being used. Businesses often have to engage specialized research teams or outside consultants to adapt models to particular workflows. Trajectory is positioning its software as an intermediary layer for that post-training work.
| Deal detail | Information |
|---|---|
| Round | Series A |
| Amount | $45 million |
| Lead investor | Sequoia |
| Other participants | Nvidia, Bessemer Venture Partners and Bain Capital Ventures |
| Valuation | $300 million |
Applications range from contracts to financial analysis
The company describes use cases in areas where businesses apply their own standards to complex work. Legal professionals could use feedback to train a model to review specific categories of contracts, while financial analysts could guide a system toward a particular method for valuing companies. Trajectory says companies including Harvey, Rogo, Decagon, Mercor and Nvidia are already using its platform.
Human input remains part of the process. Organizations can determine what constitutes good performance and use that judgment to direct subsequent improvements. Trajectory says this structure can reduce risks associated with AI systems changing their behavior without human oversight, while giving customers control over the criteria used to evaluate the models.
Post-training technology made publicly available
Trajectory is now making its post-training technology publicly available. The company’s funding will support its effort to make model improvement more accessible to non-technical users, although the extracted announcement does not disclose a timetable for additional products, revenue or employee growth. The investment places the startup in a crowded AI software market while tying its funding story to a specific enterprise need: adapting general-purpose or open-source models to the demands of individual organizations.
- Capital raised: $45 million in Series A funding
- Valuation: $300 million
- Core technology: Continual learning based on user feedback
- Named users: Harvey, Rogo, Decagon, Mercor and Nvidia
