Trajectory has opened a self-serve platform for post-training AI models on real production data. The startup was founded by former researchers from Google DeepMind, OpenAI, and Apple. It has raised about $60 million to $65 million in total.
How the platform works
The platform went live on November 2, 2026. Companies plug a lightweight SDK into their products to capture edits and corrections to AI-generated output. Those signals feed post-training cycles on a weekly basis or more often.
Funding and customers
The company raised a $15 million seed round at a $115 million valuation. A Sequoia-led round later valued the company at $300 million. Its customers include Clay, Harvey, Decagon, Mercor, and Rogo.
Product features and data
Trajectory also offers continual grading to turn user judgment into a repeatable scoring system. Its GS1 algorithm enables real-time learning from user actions. The company says client data usage follows SOC 2 standards.
User corrections are not always right. A system that learns from every edit could absorb bad habits alongside good ones. Continual grading sits at the center of the product.
Read more: techcrunch.com