AI systems can perform a wide range of tasks, but adapting them to the needs of a particular organization often calls for specialist engineers or costly consultants. Startup Trajectory is building a platform designed to help businesses improve models using feedback from the people who work with them.
The company has raised $45 million in a Series A round led by Sequoia, with participation from Nvidia, Bessemer Venture Partners and Bain Capital Ventures. The funding round values Trajectory at $300 million. Founded by three Stanford classmates with experience at Google DeepMind and Apple, the 18-person startup is now opening its post-training platform to the public.
Learning from day-to-day work
Trajectory's approach, known as continual learning, connects an AI model to a company's applications and gathers feedback on its results. That feedback is then used to retrain the model, helping it become more effective at tasks specific to the organization. Teams can begin with open-source models and define what successful performance means in their own work.
The platform is intended to make this process more accessible to professionals without technical training, including lawyers and financial analysts. Companies such as Harvey, Rogo, Decagon, Mercor and Nvidia are among its customers. Their use cases include helping models handle recurring work, such as reviewing particular kinds of contracts or applying a defined approach to company valuation.
Trajectory keeps people involved in deciding how models should improve, rather than relying on autonomous self-improvement. Its founders say this human-guided process can make specialized AI training more approachable while maintaining oversight.
By lowering the expertise and cost barriers to model adaptation, Trajectory's platform could help more organizations shape AI around the knowledge of their own teams.
