Trajectory

Trajectory launched its platform on 2 November 2026 with the argument that the people who know what good work looks like should be able to shape the model that does it, without a research lab. The same day it announced a $45M Series A led by Sequoia at a $300M valuation, with NVIDIA, Bessemer and Bain Capital Ventures taking part.

Four products folded into five lines of code

The company calls the platform the first self-serve product for post-training on production data. A customer supplies a set of tasks and a way to score them, then trains, evaluates and deploys a model in one place. Trajectory says this replaces separate sandboxes, evaluation, training and deployment tools, and that getting started takes about five lines of code, or two code changes with an existing agent harness. Training and deployment are one click each, with no sandboxes or orchestration to manage.

Teaching a model what good looks like

Two research previews point toward continual learning. Continual Grading learns from a few examples of an expert's judgment and applies those standards across production data, reducing manual labelling. Group Size 1 lets a model train on real production interactions so it keeps improving after launch. The platform also compares a customer's model with leading frontier models to show where it falls short.

Results with Harvey and NVIDIA

Builders at Harvey, Rogo, Mercor and NVIDIA are named as users, and one customer is said to have trained more than eight models for long-horizon agentic tasks in two weeks. With NVIDIA and Harvey, Trajectory post-trained Nemotron 3 Ultra on legal tasks within 24 hours of its release. It also took a Nemotron Lightning model from 0% to an 8.3% all-pass rate on Harvey's LAB benchmark, which it says matches Opus 5.5's score at 100 times lower cost and 3.8 times the speed.

A team from research labs and product studios

Trajectory says its research staff contributed to Gemini 3 at DeepMind, GPT-4o at OpenAI and Muse infrastructure at Meta Superintelligence, and that its product staff built interfaces at Stripe, Netflix and Apple. Access to the platform is through a waitlist, and the site carries field notes, documentation and a careers page.

Key facts

Launch: platform announced 2 November 2026 Funding: $45M Series A at a $300M valuation Investors: Sequoia (lead), NVIDIA, Bessemer, Bain Capital Ventures Named users: Harvey, Rogo, Mercor, NVIDIA Research previews: Continual Grading, Group Size 1 Website: trajectory.ai