Today’s AI models are brilliant on day one and stuck there. They don’t learn from their mistakes unless someone retrains them, and that work is expensive. A one-year-old startup called Trajectory just raised serious money to change that.
Quick Facts
- Company: Trajectory, founded about a year ago by three Stanford classmates
- Funding: $45 million Series A
- Valuation: $300 million
- Lead investor: Sequoia, with Nvidia, Bessemer Venture Partners and Bain Capital Ventures
- Team size: 18 people
- Source: TechCrunch (reporter Marina Temkin, November 2, 2026)
The Problem Trajectory Wants to Solve
Companies in specialized fields like law and finance usually fine-tune general models to fit their needs. That process typically requires internal research teams or costly outside consultants, often called forward-deployed engineers. The result is slow, expensive customization that many businesses can’t afford.
How Trajectory’s Platform Works
Trajectory builds around continual learning. A business starts with an open-source AI model and connects Trajectory’s platform to its applications. The platform collects feedback on how the model performs, then uses that data to retrain it so it improves at specialized tasks over time.
Co-founder Ronak Malde says non-technical experts such as lawyers and doctors can improve models themselves, almost automatically. Users define “what good looks like” for their organization, and the system turns that definition into model improvements.
The PhD Student Analogy
Malde compares foundation models to a very smart PhD student who always starts a new job. The economy, he argues, pays for someone with 30 years of experience. Trajectory aims to give models that kind of on-the-job learning.
Who’s Already Using It
Harvey and Rogo use the platform to post-train open-source models so they act more like experienced lawyers or financial analysts. Examples include learning from past tasks like reviewing a specific type of legal contract or valuing a company using a particular method. Other customers include Decagon, Mercor and Nvidia. Trajectory is now opening its post-training platform to the public.
Safer Than Self-Improving AI?
Trajectory’s approach has some similarities to recursive self-improvement, the idea that AI systems improve their own capabilities. Co-founder Michael Elabd argues that keeping humans in the loop makes it safer, since fully autonomous self-improvement could lead to unintended or even dangerous results.
A Crowded Field
Trajectory isn’t alone. Competitors include Thinking Machines Lab, founded by former OpenAI CTO Mira Murati and behind the fine-tuning platform Tinker. Applied Compute, valued at $3.25 billion last month, and Prime Intellect, which helps enterprises build and train their own agents, also compete in this space.
The Team Behind It
Malde was among researchers Google DeepMind acqui-hired from Windsurf. Elabd also worked at DeepMind, and co-founder Arjun Karanam previously worked on generative foundation models at Apple. The startup has hired talent from OpenAI, Stripe and Netflix.
Why Design Matters
Karanam says Trajectory built an interface that makes post-training accessible to non-technical customers. For a company aiming to serve doctors and lawyers rather than machine learning engineers, that design focus is central.
What It Means for Businesses
- Lower cost: Specialized AI could become cheaper to build.
- Faster deployment: Domain experts can guide training directly.
- More control: Companies can start with open-source models instead of locking into one provider.
What to Watch Next
Look for the public launch’s uptake, new enterprise customers and how Trajectory competes with better-funded rivals.
Source: TechCrunch