Technology
Trajectory Launches Self-Serve AI Platform for Real-Time Model Training
Trajectory, a startup founded by former researchers from Google DeepMind, OpenAI, and Apple, has officially launched a self-serve platform designed for post-training AI models, enabling businesses to utilize real production data for continuous model improvement. The platform went live on November 2, 2026, marking a significant shift from working closely with a select group of partners to offering services to any qualifying business.
The platform operates through a lightweight software development kit (SDK) that companies can integrate into their products. This SDK captures user interactions, such as edits and corrections made to AI-generated outputs, which then inform post-training cycles aimed at refining the model's performance. Trajectory emphasizes that these cycles can occur weekly or even more frequently, allowing for rapid adjustments based on user feedback.
One of the standout features of the platform is its continual grading system, which seeks to translate user judgments into a repeatable scoring mechanism that the AI can learn from. Additionally, the GS1 algorithm enables real-time learning from user actions, enhancing the model's adaptability.
Trajectory was founded by Ronak Malde, Arjun Karanam, and Michael Elabd, and has quickly gained traction in the AI space. The company raised a total of approximately $60-65 million in funding, including a notable $40 million Series A round led by Sequoia Capital, which valued the company at $300 million.
As the platform opens up to a broader audience, it raises important considerations regarding data privacy and the quality of user corrections. Trajectory's commitment to adhering to SOC 2 standards reflects its awareness of these challenges, particularly as enterprises express concerns over the governance of customer data.
FAQ
What is the purpose of Trajectory's self-serve AI platform?
Trajectory's self-serve AI platform is designed for post-training AI models, allowing businesses to utilize real production data for continuous model improvement and refinement.
How does the platform integrate with existing products?
The platform operates through a lightweight software development kit (SDK) that companies can easily integrate into their products to capture user interactions and feedback.
What is the significance of the continual grading system?
The continual grading system translates user judgments into a repeatable scoring mechanism, enabling the AI to learn from user feedback and improve its performance over time.
What funding has Trajectory received?
Trajectory has raised approximately $60-65 million in funding, including a $40 million Series A round led by Sequoia Capital, which valued the company at $300 million.
How does Trajectory address data privacy concerns?
Trajectory is committed to adhering to SOC 2 standards to ensure data privacy and governance, addressing concerns that enterprises may have regarding the handling of customer data.