Every year, academic researchers create algorithms and craft lengthy papers about their work. A few people might read the papers, but for the most part, those algorithms get forgotten. Well, that could be changing.
A startup called Algorithmia has constructed a place for countless algorithms to live, allowing developers to try them out and find the most effective ones. Investors liked the concept enough to put $2.4 million in seed money behind the startup.
It’s one of these ideas that help multiple kinds of people. Developers get more choices. But at the same time, people who cook up algorithms can focus on that type of work entirely, so they don’t have to take jobs where they do that for only a small fraction of the time and do boring work for the rest of the day. They can send their work to Algorithmia and then sit back and watch the money roll in.
“We’re saying you can be an algorithm developer 24-7,” Diego Oppenheimer, Algorithmia’s chief executive and a co-founder, said in an interview with VentureBeat.
The startup’s vision bears some resemblance to the marketplaces in existence today for business applications and services — take public-cloud provider Amazon Web Services’ marketplace, the Salesforce.com App Exchange, or 99designs’ design marketplace as examples. And of course, consumers can go to marketplaces of different providers. Uber and Airbnb come to mind. Many of these communities have taken off, but to date such a virtual bazaar for algorithms appears to be new.
And it’s a good time for such technology to come about. Companies big and small are hiring data scientists who can explore data hanging around or start working with product teams to design software that can produce data. Data scientists can then pick up on trends and anomalies and make suggestions about company strategy and product development.
But a few algorithms end up getting used far more than others. Perhaps a way to test algorithms could yield more powerful results for software companies, or just applications that provide a better user experience.
And the tool could save time for data scientists. There’s no need to spend hours reading a paper and then tinkering with an algorithm to get it ready to handle data.
“Every algorithm inside Algorithmia is always live,” Oppenheimer said. “The code is always live and running.
An algorithm developer comes in, they submit their code, compile it, and distribute it to our compute nodes.”
Then people can tap in to the algorithms through Algorithmia’s application programming interface, or API.
Algorithmia charges a base price for its service and offers a certain amount of usage of the API for that price. Customers can pay for blocks of additional API usage, and Algorithmia keeps track of which algorithms get used. That’s how it knows how much to pay the developers behind specific algorithms, Oppenheimer said.
“It’s a crowdsourced approach to artificial intelligence,” he said.
Madrona Venture Group led the round in Algorithmia. Deep Fork Capital, Rakuten Ventures, Oren Etzioni, and Charles Fitzgerald also participated.
Based in Seattle, the startup employs four people, and plans call for that number to shoot up to eight.