Labelbox, the leading training data platform for enterprise machine learning applications, announced the close of a $25 million Series B funding round led by Andreessen Horowitz with General Partner Peter Levine joining the Labelbox board of directors. Previous investors First Round Capital, Gradient Ventures (Google’s AI-focused venture fund) and Kleiner Perkins also participated. To date, Labelbox has raised $39 million in venture funding.
Labelbox offers a training data platform for machine learning teams to build real-world artificial intelligence. The platform consists of label editor tools, batch & real-time labeling workflows, collaboration, quality review, analytics, and an optional, fully managed and dedicated labeling workforce. With state-of-the-art algorithms available for free and AI computing costs drastically reducing, high-quality labeled training data is the most valuable asset for enterprises adopting supervised learning solutions. In less than two years, Labelbox has become a foundational piece of infrastructure for more than one hundred companies that are operating production-grade AI today.
The world is very quickly moving into a new paradigm where domain experts directly teach artificial intelligence about the skilled work they want it to do. This new paradigm is often called data-centric programming or Software 2.0. Labelbox aims to become the de-facto training data platform for Software 2.0, similar to what GitHub is for Software 1.0. The Labelbox platform focuses today on computer vision applications but can handle all forms of data.
Labelbox is defining a new category: training data platform. The Labelbox product provides scalable tools, workflows, collaboration, quality review, and automation to its customers on a unified platform. Customers are taking control of their data and using various out-of-the-box labeling tools and workflows to reliably build products and services with AI. In addition, Labelbox is accelerating new AI development across the enterprise because it serves as the company’s central hub for all training data.
“If GitHub has become the platform for managing and developing software code, then Labelbox has the potential to fill a similar role for data in the AI/ML world,” said Peter Levine, general partner at Andreessen Horowitz. “We see Labelbox becoming the single source of truth for defining, storing, and accessing training data across an entire organization. We are thrilled to partner with company founders Manu, Brian, Dan, and the team to help them realize their vision.”
Sharma saw the opportunity to start Labelbox while working at Planet Labs, the California company that photographs the earth from space and analyzes more than seven terabytes of images daily in various ways. Engineers at Planet Labs needed a software platform to create and manage training data at scale and there was no off-the-shelf solution, so they built one themselves. “A lot of this tooling was being built from scratch,” said Sharma. “It seemed crazy to us that data scientists were building core infrastructure in order to get started with AI.”
Sharma and his Labelbox co-founders realized Planet Labs wasn’t alone. Every company building supervised learning models faced the same problem. So they started Labelbox, an alternative to turning data over to ‘black box’ labeling services where companies have no control.
Labelbox is currently being used by industries as diverse as agriculture (where it is used to identify weeds in fields, for example), to insurance (to spot risks), to sports analysis (to track the spin on balls or the actions of players on the field), to healthcare (identifying tumors or cells). With the new funding, Labelbox intends to accelerate its roadmap and go to market. As supervised learning transforms economies around the world, the opportunities for Labelbox are seemingly endless.