Labelbox

Labelbox is a San Francisco company that began as a data labeling platform and now sells training environments and evaluation data to AI developers, alongside software agents for enterprise teams. On 14 October 2026 it announced a Series E of more than $100 million, combining equity and a credit facility, at a $1.25 billion pre-money valuation, led by Infinitum Partners. The company says it is profitable in 2026.

What it does

Labelbox builds two main product lines. Horizon provides reinforcement learning environments and evaluations used in post-training, covering reasoning, tool use and computer use. It includes a simulator that generates enterprise applications and business data for agents to work in, plus expert preference labels. Recursion, or Recursion Managed Agents, runs background agents that start from schedules, alerts, commits or API calls and keep working until a task is finished, with sandboxes, stored credentials, memory and rubric-based grading. The site also lists Terra, a robotics data offering, and Alignerr, a network of skill-tested experts who write rubrics and judge outputs. It publishes research benchmarks as well.

Who it serves

The company's primary audience is frontier AI laboratories that post-train and evaluate models. Enterprise engineering, research and operations teams are the audience for Recursion. Labelbox reports more than 400 customers, and its site shows logos from media, medical technology, retail and consumer firms. It lists SOC 2 Type II, GDPR, CCPA, NIST 800-171 and PCI DSS among its compliance credentials.

Key facts

Founded: 2018, by Manu Sharma and co-founders Headquarters: San Francisco, California Funding: Series E of more than $100 million (equity and credit facility), October 2026, led by Infinitum Partners, at $1.25B pre-money Customers: 400+ Products: Horizon, Recursion Managed Agents, Terra, Alignerr