Tue · Jul 26

Scale is an API for human labor

Scale is an API that lets companies outsource manual tasks. With one line of code, companies can employ a human to do on-demand tasks like data extraction, calls, or categorization. Founders, Lucy Guo and Alexandr Wang, sat down with us to tell us more about Scale.

What YC Liked About Scale:

"The great thing about Scale is that it is taking advantage of an opportunity that is hiding in plain sight. Because Mechanical Turk has been around for a long time, no one has been building startups that could compete with it. Meanwhile, Mechanical Turk has degenerated - becoming overridden with spam and heavily restricted in use cases. More of our companies are hiring their own freelancers rather than using a service to accomplish tasks, which shows that there is a huge opportunity for a company like Scale to do human outsourcing right."
- Jared Friedman, Partner, Y Combinator

YC : What problem are you solving?

Lucy: We’re building the simplest way to outsource a human task.

Every company still has core processes that are powered by humans, stuff like content moderation, data extraction or scheduling appointments. Currently, companies have to hire, train, and manage outsourced teams to handle these tasks. Scale takes away the pain of setting up and managing these outsourced teams.

With one line of code, a company can have tasks completed on-demand. That means the company can focus on building its core business.

YC : Tell us about how it works? And why is this better than something like Mechanical Turk?

Alexandr: Companies just have to put in one line of code to outsource their tasks through Scale (you can read the docs here).

On our end, we hire contractors who do the tasks for us. Mechanical Turk is 40% spam because there’s no formal vetting process to test the competencies of their freelancers. If you use Mechanical Turk, you need to do your own quality control, sending tasks to multiple workers and picking the most common response. This creates a lot of additional work for companies.

With Scale, we do quality control on the contractors so you know you can trust your first answer. Unlike Mechanical Turk, which is designed for batch jobs that get responses in hours, our contractors provide responses within minutes and are suitable for on-demand tasks. We also provide a broader range of services than Mechanical Turk does. For example, we can do phone calling.

We’re basically the simplest API available to do this kind of work.

YC : Why is now the time for this?

Lucy: There’s a big trend right now of human-powered apps and chatbots. Companies are still trying trying to gather enough data for their machine learning algorithms, so they’re using humans to do a lot of the tasks they claim are being performed by “AI.”

Large companies like Google and Facebook have long had large teams of workers they use for these kinds of tasks. More and more startups are finding they need the same pools of quality workers, but it doesn’t make sense for every startup to build out their own team. Scale gives every company the access to human workers that Google has.

YC : How did you start working on this particular problem?

Lucy: Prior to this, we worked at Quora and Snapchat, and they rely heavily on manual content moderation. When someone flags an image, it gets sent to another person who checks whether the content needs to be removed or not. It’s a very manual process and we kind of wished there was an API to just do all of it for us.

YC : What’s the big vision for Scale?

Alexandr: Long-term we want to power any human-powered process for any company. A lot of these things will eventually be automated out by AI/machine learning. By looking at all the data we get from companies, we’ll be able to automate 90% of the request. The remaining 10% will be handled by humans for things that fall into grey-areas to maintain high quality, and we’ll be there to power that as well.

Sign up for weekly recaps of The Macro.