Welcome, Scale!
Welcome, Scale!
We are thrilled to be partnering with Scale, alongside our friends at Founders Fund, Thrive, Coatue, Index, Accel, and more.
The Spark partnership believes strongly in the transformational impact that machine learning and artificial intelligence will have over time, and we have been fortunate to partner with companies on the forefront of these emerging technologies in the past. The applications powered by advancements in machine intelligence (across computer vision, natural language processing, autonomous systems, process automation, and more) will undoubtedly transform myriad end-markets over the next couple of decades and beyond.
While much has been written about the potential impact of computer vision on the automotive sector (through the development of AVs), we believe that applications leveraging computer vision will also transform manufacturing, defense, security, healthcare, construction, retail, agriculture, and more. At the same time, improvements in NLP should allow for more effective chat-based applications and content moderation on social platforms, while process automation software will drive hundreds of $Bs of efficiencies in the enterprise. The potential implications of these emerging technologies are still unknown — and likely underestimated.
That said, there are still bottlenecks in the AI development cycle that need to be solved to allow for the democratization of access to machine learning at scale. In order for machine learning systems to create accurate generalizations, they must be trained on data. More advanced forms of machine learning, like deep learning neural networks, require an immense amount of data to create models with desired levels of accuracy — and continue to get better with more data, even at massive scale. With this in mind, we believe that access to training data — not better algorithms — will determine the winners and losers in an AI-driven world. For the most part, these applications will not be improved by developers writing better code — instead, they will be improved by training on more/better data, learning from the results, and repeating. The applications that have unfair access to data will have unfair advantages, as the AI development cycle creates a natural flywheel. Much of the data generated by applications, however, is unstructured and relatively unusable as-generated. Models need to be trained on data that is clean, accurate, complete, and well-labeled — as you’d expect, ML development tends to follow the rule of garbage-in, garbage-out.
Scale is accelerating the development of AI applications by providing an API-driven platform for reliable, high-quality labeled training data. Customers send Scale sensor data on an ongoing basis via API call, and through a combination of (i) machine learning automation, (ii) tools with optimized UX, and (iii) scaled human insight, Scale returns accurate ground truth data. Scale has built both technology and operational stacks that allow them to go-to market with a scalable, reliable, high-quality, and cost-effective data labeling engine, allowing developers of AI applications to outsource an ops-intensive part of their development cycle (that is not core to the IP they are developing) to a best-in-class vendor. Put simply, Scale is taking one of the least glamorous — but most critical — parts of the machine learning problem set and solving it for every other AI team.
While much of Scale’s early traction has come from the automotive sector, the need for high-quality training data will be pervasive across any AI end-market, and customers are already leveraging Scale’s labeling engine for NLP use-cases and CV use-cases in robotics, drones, retail, security and more. In addition, Scale is excited to participate more broadly in its customers’ AI development cycles over time, as they push forward their mission of accelerating the development of AI-driven applications globally.
We have gotten to know Alex and the Scale team well over the last couple of years and are convinced they are the right team to solve this problem (and they are hiring!). We are excited to be partners on the journey.
