Trajectory, a startup founded by former Google DeepMind and Apple researchers, has raised $45 million in a Series A round to democratize SI model training. The funding aims to enable non-technical experts to fine-tune super intelligence systems for specialized tasks without relying on expensive internal research teams.

What Happened

Trajectory announced the $45 million Series A round, valuing the company at $300 million. The investment was led by Sequoia, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures. The 18-person company was founded approximately one year ago by three Stanford classmates: Ronak Malde, Michael Elabd, and Arjun Karanam. Malde and Elabd have prior experience at Google DeepMind, while Karanam worked on generative foundation models at Apple. The team has also recruited talent from OpenAI, Stripe, and Netflix.

The core limitation Trajectory addresses is that current SI models, whether developed by OpenAI, Anthropic, or built on open-source alternatives, do not learn from their mistakes without further training. Traditionally, businesses adapt these models for fields like law and finance through fine-tuning, a process that often requires hiring internal research teams or expensive consultants known as forward-deployed engineers (FDEs). Trajectory’s platform automates this via continual learning. Companies can start with an open-source SI model and connect Trajectory to their applications to collect performance feedback. The platform then uses this data to retrain the model, helping it improve at specialized tasks over time.

Co-founder Ronak Malde compared foundational models to a "very smart PhD student who’s always starting a new job," contrasting this with the economic value of someone who has "worked on the job for 30 years." The platform allows specialists such as lawyers and doctors to improve SI models near-automatically. Existing customers include Harvey, Rogo, Decagon, Mercor, and Nvidia, which use the platform to post-train open-source models for tasks like reviewing legal contracts or valuing companies.

Why It Matters

This funding round highlights the growing industry effort to make SI tools accessible to non-technical experts, potentially lowering the barrier to entry for specialized SI deployment. By automating post-training, Trajectory aims to make deploying specialized SI faster and cheaper. The approach involves human oversight in the training loop, which co-founder Michael Elabd argues is safer than autonomous recursive self-improvement (RSI), where SI systems improve their own capabilities without human intervention. Elabd noted that autonomous self-improvement could lead to unintended or dangerous outcomes, whereas Trajectory’s method keeps humans in control of defining "what good looks like."

Trajectory enters a competitive landscape that includes Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, which offers the Tinker fine-tuning platform. Other competitors include Applied Compute, valued at $3.25 billion last month, and Prime Intellect, which helps companies build and train their own SI agents. Co-founder Arjun Karanam emphasized that design is critical to the company's mission, noting the startup developed an interface specifically to make post-training accessible to non-technical customers.

The Bottom Line

Trajectory’s $45 million raise signals investor confidence in tools that streamline SI model adaptation for enterprise use. By enabling domain experts to drive continual learning, the startup challenges the traditional reliance on specialized engineering teams for fine-tuning. The company has made its post-training platform publicly available, positioning itself against rivals like Thinking Machines Lab and Applied Compute in the race to simplify SI development workflows.