1) Tell us about your background and how/why you started the company.
I co-founded Labelbox with Brian Reiger and Dan Rasmuson in 2018. The idea came from insights we uncovered while working at our respective companies at the time. I was at Planet Labs, one of the leading geospatial satellite imaging companies. Planet Labs operates like a Starlink constellation but with satellites taking images.
At the time, deep learning was taking off, and there was a lot of excitement around self-driving cars. However, we saw a broader opportunity: deep learning models were going to power many facets of life. The only way to create those models at the time was by labeling data with human input. We saw an opportunity to build a product for this and initially, we thought Labelbox would be a smaller scale business, not a venture-backed company. This was our third business together with Brian.
We moved to a low-cost place outside the Bay Area and built a product on nights and weekends, launching it on Reddit. It took off quickly. Every week, more people signed up, and we were excited about solving their problems and learning about new ones. Within a couple of months, we were generating revenue and had a list of features our customers wanted. At that point, we knew we were onto something and decided to raise funding and scale.
2) What are the key problems you are addressing and how you are delivering value to customers through software workflows?
The key problem that Labelbox solves for our customers is producing high-quality data. Developing AI models, whether it’s frontier models like LLMs and multimodal models or task-specific models for mainstream enterprises, requires two key things: compute and data. Labelbox is the data provider and we serve essentially as a data factory that produces the data needed for these models.
We specialize in creating data that is vetted and curated by human expertise. Labelbox offers tools and dedicated software for customers to do it themselves, and we provide an end-to-end service. All of these capabilities are part of the same platform and this is what makes Labelbox unique. Other players might provide labels in a CSV file or an API call, but it’s up to the customer to vet the data quality. With Labelbox, we offer all the tools needed to gain full transparency and control over the process to consistently deliver high-quality data.
For example, teams looking to improve quality can make real-time comments, track label quality, and access performance dashboards. Both the labelers and our customers use the same tools on the same platform. This integrated approach streamlines the process and makes life easier for our customers. We see human evaluations (evals) as part of the same problem as data labeling. Whether it’s instruction fine-tuning (SFT) or preference ranking, it’s all about asking humans for feedback. This feedback can take various forms, from supervised experiences to semi-supervised techniques like clustering or pre-labeling with foundation models.
3) What have you enjoyed or what do you look forward to about working with Tau Ventures?
We’ve really enjoyed working with the entire Tau Ventures team whether it’s getting their insights on specifical verticals for our GTM efforts, or joining their in-person events. I’ve especially enjoyed presenting at Tau Venture’s Annual Day, which fosters a strong sense of community and allows founders to network, share experiences, and learn from each other in person.
