Eight years ago, Amazon revolutionized the way companies could draw on human labor by creating its Mechanical Turk platform. With "MTurk," as it is sometimes known, workers could be delivered the same way Amazon sold computing resources through its AWS division: on demand, via the cloud, and in a way that was capable of scaling up and down with ease. The only drawback was that Mturk was largely limited to "simple human intelligence tasks," or as critics called them, "SHITS."

In recent years, however, there has been a push to evolve the capabilities of the field known as crowd computing: a mix of human and machine intelligence capable of executing more complex tasks. The newest entrant to the field is New York startup Fancy Hands, which offers on-demand personal assistants. Fancy Hands recently rolled out a "human API" that allows developers to easily add an army of human workers to any mobile or web app.

"Mechanical Turk is really great but a huge majority of the tasks on there are simple things like deciding if video content is not safe for work or not," says Ted Roden, Fancy Hands' founder and CEO. "We wanted to do something a few steps above that." So far, Fancy Hands has been baked into apps from big brands, so that if a customer tweets a request through the service, a real human can go out and fulfill it. "It might be really complex to design an app that books a restaurant reservation for you, but a human can easily accomplish that."

"API-enabled worker communities are the future of labor," says Max Yankelevich, the chief architect at Crowd Computing Systems, a competitor of MTurk and the Fancy Hands API. "Software can leverage these ‘hybrid crowds’ to distribute customers' work among qualified workers in various communities." By hybrid, Yankelevich means a combination of human and machine intelligence. "That's what crowd computing is all about — separation of labor providers and software-platform providers."

By combining complex offline work with the kind of simple microtasks laborers can do online, crowd computing can produce more interesting results. "The next step in evolution is to wire tasks together," says Yankelevich. "For example, engage a worker to take a picture of products on a store shelf and then have a different crowd worker analyze those pictures for compliance with the brand's marketing rules for shelf display."

There are around 15,000 people in Fancy Hands’ on-demand labor pool, with an average of 1,000 actively working on tasks each day. Some critics have accused this kind of startup of replacing good full-time jobs with piecemeal work. "Our bread-and-butter workers are folks from small-town America struggling to find work who would rather do this full-time than work at the local gas station," says Roden. "So I don’t believe it's exploitative." Piecemeal workers don't get health insurance or a minimum wage, but they do set their own price for each task.

Fancy Hands has an assortment of plans offering between five and twenty-five chores per month. For developers looking to build Fancy Hands' workforce into their apps, the API is free, and they decide what they are willing to pay per task. Roden says he hopes to improve the service over the next year by focusing on predictive intelligence. "The ultimate goal would be to help build an app that anticipates your needs, like Google Now, but that actually does the hard work of buying your airline ticket or calling the car service."