Trajectory Secures $45M to Simplify AI Model Fine Tuning

Trajectory has raised $45 million in a Series A funding round, reaching a valuation of $300 million to build software that automates artificial intelligence training for non-experts. The startup, reported by TechCrunch, is led by founders Ronak Malde, Michael Elabd, and Arjun Karanam, who previously held engineering roles at Google DeepMind and Apple.

Automating Continual Learning for Enterprises

The platform allows businesses to connect open source models to their internal applications, enabling the systems to learn from ongoing feedback loops. Instead of hiring specialized forward-deployed engineers to manage fine-tuning, domain experts like lawyers or financial analysts can effectively retrain models on proprietary data. The Scale Report understands that by keeping humans in the loop, the team avoids the potential safety pitfalls associated with fully autonomous recursive self-improvement.

Competitive Landscape and Market Adoption

Early adopters of the technology include Harvey, Rogo, Decagon, Mercor, and Nvidia. The company is positioning itself against other emerging infrastructure players such as Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, as well as Applied Compute and Prime Intellect.

This shift toward automated post-training represents a critical pivot in enterprise AI adoption. Many foundational models, such as those from Anthropic, often struggle to retain domain-specific institutional knowledge. By reducing the reliance on high-cost consulting teams, Trajectory aims to lower the barrier for specialized sectors to maintain AI agents that perform consistently over long time horizons.