Behind every AI, there’s a network of people whose job it is to sift through millions of data sets, annotate the information and feed it back into the system.
These data labelers curate the information that teaches an AI to flip burgers, drive cars and streamline office tasks. They are the eyes and ears of an AI before it can see and hear. It’s a critical job, but also a time-intensive one that often requires large teams and thousands of work hours to do right.
For most companies trying to create their own AI system, it’s also their biggest hurdle. While they may have the data, most don’t have employees to sift through it and make it usable. Alegion hopes to fill that role for companies, offering both manpower and a platform to help process the data.
“The whole point of AI is to mimic human judgement. We combine tech with humans to provide scaled high quality data.”
On Thursday, the Austin-based company raised $12 million in its second Series A to build on those efforts. With data labeling poised to become a billion dollar industry in the next four years, the round has Alegion poised to take advantage, Gates said.
“This allows us to continue to build our software and capabilities as our market matures,” he said. “For the first time we’re scaling our go-to-market team and building out sales organization.”
Gates initially launched Alegion in 2012 as a crowdsourcing tool. However, three years in, he noticed companies asking for help processing data for machine learning. With AI starting to gain traction as a tool for all industries, Gates realized there was an opening for a company that could help others tag data to train their AI models.
That prompted the shift to focus on data labeling.
“Data science teams usually come to us frustrated because they can’t label enough data to move the needle on their models,” Gates said. “We’re able to do that, and it’s very valuable to them. It’s scale and accuracy, and you have to have both at the same time.”
To do so, Gates combined Alegion’s crowdsourcing capabilities with a machine learning tool to assist in data labeling.
Through Alegion, companies can submit their raw data for a labeling task like identifying strawberries in a picture, for example. That data, which could be one photo or thousands, is then divided up and distributed to thousands of pre-screened workers whose job it is to box or label each strawberry they see in an image to make the data useful. The platform then uses machine learning to support the labelers and ensure accurate work.
“The whole point of AI is to mimic human judgement,” Gates said. “We combine tech with humans to provide scaled high quality data.”
The company has started to see an increase in demand for its data labeling services. Alegion has tripled its number of proofs of concepts and projects from quarter one in the last quarter, Gates said.
This round will help the company build out its sales and marketing team and improve its machine learning platform to capitalize on those opportunities. Already at 58 employees in Austin, the company plans to grow to over 100 by next year.
“This bought us time for the market to mature and allows us to perfect our tools and software for when the market does mature,” Gates said.
RHS Investments led the round.
