Graeme Dawes - Fotolia
Meet James, the new QA in the next cube. He's highly trained, scarily efficient, and he's happy with only $5 an hour. He's also a virtual robot.
James is the just-released brainchild of cloud services provider dinCloud, which created the virtual robot to work with existing IT or software development monitoring systems to track and report on customer experience. It's a dramatic change in direction for dinCloud, and it could presage an even greater change in the software testing world.
At a time when the desire for DevOps has meant a move to automated testing by companies large and small, the idea of a service like James is likely to send at least a shiver throughout the quality assurance (QA) industry. A study done late last year by London-based executive placement firm Harvey Nash showed 67% of software testers expect they'll be automated out of a job in 10 years. And James isn't going to be the only new QA tool based on artificial intelligence heading into the software development space in the next few years.
Longtime tester and test manager Gerie Owen, who is also a regular contributor to TechTarget, had a "wowsa" moment when told about James. But she was quick to stress the need for human involvement in software testing.
"These are not QA people; they are computer applications programmed with data," she said. "Customer experience is a very ambiguous problem domain, and the rules change based on the application and target audience. I would imagine that the data this robot is trained on also has to be ambiguous. While I think that such a tool could be a valuable assistant, it almost certainly has inherent biases that will influence its results."
DinCloud's goal for new QA James is to focus on just what a user is experiencing, whether the virtual robot is monitoring an internal IT system or looking at a just-released app update, said Ali Din, company general manager and chief marketing officer. "When an end user has a cart and is checking out, the system transfers momentarily to the credit card payment gateway, and the business doesn't control that gateway or see any user issue," he said. "With James coming in as the user, he can see what the error message is and capture and report it across multiple systems."
Din said James can also be a good source of information about updates by checking performance after an update is released and comparing it with earlier data. "This way, a company doesn't have to wait and have users tell them what the problem is. This is a way to be more proactive."
To put it another way, James functions as a new QA, with a total focus on the business process. "James doesn't care about what the app does, but concentrates on the business process steps," Din said. "This is the stuff that if one tiny link is broken, nothing works."
And Din said James is also relatively low-maintenance. "James gets installed on a workstation dedicated to him -- just like if you were hiring a QA pro, you'd give him a workstation." He works from a list of tasks independently and communicates using the company's internal network, or via email or text. He has the capability to escalate issues if necessary.
And James' salary works out to about $5 an hour -- James' list price is $3,600 a month, Din said, but noted that since the company works with a lot of channel partners, there's a good chance for some discounting at some point in the future.
Price aside, skepticism remains. Matthew Heusser, a longtime tester, consultant and managing director of Excelon Development -- as well as a regular TechTarget contributor -- said James "sounds like a chatbot, and my experience with chatbots has not been very good." His take: Wait and see if the company can work out the technical challenges, and then re-evaluate in a year or so.
Owen is less sanguine. "If a development team comes up with an entirely new concept of customer experience that takes it to the next domain, the robot hasn't been trained on that and won't be able to easily adapt -- like a human could," she said. But even more troubling to her is the issue of oversight: "Who will watch the watchmen?"
Everyone is starting to take QA more seriously -- here's why
Why scriptless test automation is going to be the next big thing
Natural language is going to be a tester's new BFF