Key points
- Trajectory raised a $45 million Series A at a $300 million valuation, led by Sequoia, with Nvidia, Bessemer and Bain Capital Ventures participating.
- Its post-training platform is now publicly available; Harvey, Rogo, Decagon, Mercor and Nvidia are named as customers.
- Teams start from an open source model, connect it to their apps, and the platform retrains it using collected feedback.
- Rivals named include Thinking Machines Lab (Tinker), Applied Compute and Prime Intellect.
Why it matters for marketers: Specialized AI that learns from expert feedback without a research team could let content and marketing groups tune models to their own standards of quality.
Trajectory, a young startup with three co-founders who met at Stanford, announced a $45 million Series A at a $300 million valuation. Sequoia led the round, and Nvidia, Bessemer Venture Partners and Bain Capital Ventures took part. The company has 18 employees and is releasing its post-training platform to the public.
The product targets a known gap: general AI models do not learn from their errors unless they are trained again. Companies in fields like law and finance usually close that gap by fine-tuning, which tends to need internal researchers or costly outside engineers. Trajectory wants to automate the work. A business begins with an open source model, links the platform to its applications, and lets it gather feedback that is then used to retrain the model.
Co-founder Ronak Malde told TechCrunch that lawyers, doctors and other specialists can improve models almost automatically on their own. Harvey and Rogo already use the platform to make models behave more like seasoned lawyers or analysts. Co-founder Michael Elabd argues that keeping people in the training loop is safer than fully autonomous self-improvement.
For marketers and content teams, the appeal is control over what counts as good output. If the people who know the brand voice, compliance rules or editorial standards can shape a model directly, specialized tools may arrive faster and cost less. The market is crowded, though. Trajectory names Thinking Machines Lab, Applied Compute and Prime Intellect as competitors, so teams should compare options and test how well feedback-driven retraining holds up on real workflows before committing.
Sources
This brief is our own summary. Read the full reporting at the source.
