Trajectory raises $45M to launch platform for continual AI model training by nontechnical users
Series A values the 18-person startup at $300M; Sequoia leads with Nvidia, Bessemer, and Bain participating.
1 source · cross-referenced
- Trajectory emerged from stealth with an $45M Series A at a $300M valuation led by Sequoia, with Nvidia, Bessemer Venture Partners, and Bain Capital Ventures participating.
- The platform lets businesses start from open-source models and iteratively improve them using domain-specific feedback without hiring specialized engineers.
- Customers like Harvey, Rogo, Decagon, Mercor, and Nvidia are already using the post-training platform to refine models for legal and financial workflows.
Trajectory, an 18-person startup founded about a year ago by three Stanford classmates with prior roles at Google DeepMind and Apple, emerged from stealth with a $45 million Series A at a $300 million valuation. The round was led by Sequoia, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures.
The company’s platform targets a gap in current AI workflows: models rarely learn from mistakes without additional training, and businesses often rely on expensive internal research teams or consultants to fine-tune models for specialized domains like law or finance. Trajectory says its platform automates continual learning, allowing organizations to start from open-source models and iteratively improve them using feedback collected directly from their applications.
Trajectory co-founder Ronak Malde likened foundational models to “a very smart PhD student who’s always starting a new job,” emphasizing the need for domain-specific experience over time. The startup’s approach centers on continual learning, where feedback from real-world usage is used to retrain models, helping them improve at specialized tasks without requiring deep technical expertise from end users.
Customers already using the platform include Harvey and Rogo, which apply it to post-train open-source models for legal and financial workflows, such as reviewing contracts or valuing companies. Trajectory also lists Decagon, Mercor, and Nvidia among its customers. The company announced that its post-training platform is now publicly available.
Trajectory positions its method as distinct from recursive self-improvement, arguing that human involvement in the training loop makes the process safer and more controllable. Co-founder Michael Elabd contrasted this with autonomous self-improvement, which he suggested could lead to unintended or dangerous outcomes.
The startup identifies Thinking Machines Lab—founded by former OpenAI CTO Mira Murati and maker of the fine-tuning platform Tinker—as a direct competitor, along with Applied Compute and Prime Intellect, which also focus on post-training and agent-building platforms.
- Nov 11, 2026 · Hugging Face
Hugging Face proposes token pooling-aware training to improve compressed multi-vector retrieval
Trust79 - Nov 11, 2026 · TechCrunch — AI
Fractile raises $720M Series C to develop AI chips and compiler stack
Trust79 - Nov 11, 2026 · The Verge — AI
Google’s Tabletop AI experiment lets users build custom collaborative apps via waitlist
Trust79