Pretty much every company, to some extent, has to power their algorithms with data and machine learning if theyre going to be competitive with anyone else in their space. The challenge there, though, is that not everyone can have the massive computational power and data of a company like Google especially not startups.
Siddharth Mall and his co-founders started Playment in India based on that bet. Playment allows companies to send over sets of training data that need some kind of light analysis, and then divvies it up across a large on-demand workforce that can resolve those tasks on their mobile devices. That, in turn, feeds back additional data to the companies to help refine their algorithms whether thats visual search, quality checks of recommendation engines.
Playment users can do a lot of the testing and actions on their mobile devices. So it could be as simple as making small quality checks while theyre commuting or drawing bounding boxes for visual recognition (like below), giving the option of doing these kinds of tasks anywhere. So users can do as many or as few tasks as theyre looking for whether its just for some additional early cash or pushing it toward more of a primary income amount, Mall said.
Companies pass the data off to Playment through a set of APIs, and then Playment creates a workflow for its workforce. Then those workflows are passed off to their workforce, and users can pick up whatever tasks they want that are available. Mall said that with that approach the company can return the requests in as little as a few minutes, though it would depend on the complexity of the task. One possibility, with this approach, wouldalso besplitting complicated tasks into multiple workflows and divvying them out to multiple people.
You will need to enhance your algorithmby training it on multiple fronts, Mall said. Today you need to count pedestrians, tomorrow you probably need to read signs. The day after that,you need more human judgment calls. Obviously its not just one particular tasksthroughout. But an algorithmis only as good as the data that trains them.
The obvious use case here is in validating product reviews for companies like Flipkart and other e-commerce companies. By offloading the process of checking product reviews to a wide array of people instead of a small quality control team, e-commerce sites could post more reviews, more quickly, and create a more accurately determine the best products available for recommendations.
The goal, then, is to take the load off of companies like Flipkart and others that have to do these kinds of quality checks in order to keep things moving on their sites. That can relieve financial pressure, and offer some alternative income for other individuals,Mall said though, of course, that means fewer jobs that are available at Flipkart. But, like Uber, the flexibility of something like Playment is what is going to make it attractive, he said.
The obvious competitor is Mechanical Turk from Amazon, which can also accomplish these kinds of small tasks. Mall said the biggest question hed regularly get from investors and potential customers is whether or not Playment can ensure that the quality of results is high especially for the volume of requests. Thats the goal Playment is gunning for, and Mall hopes that will help it be more successful than Mechanical Turk.
There is, of course, probably a more difficult side to all of this. These kinds of taskscan be mundane, and its hard to imagine that this is something that would psychologically fulfill someone. Mall said the tool offers people an opportunity to make more than $100 USD extra each month doing these tasks. And again, the play here is the flexibility that can keep people working on the service.
We give them the flexibility to work at home, or wherever they wish, becauseits on mobile, Mall said. You could pay for coffee youre standing in line for by doing simple tasks. Its not like Mechanical Turkwhere youre forced to sit in a place and do the same things. Its flexible and open and you can do whatever you want, and you can earn money as much as you can earn in a normal daily job.