Laurent - stock.adobe.com
IT services and solutions provider Mphasis claims to have developed an internal artificial intelligence (AI) tool, which means it now takes its software developers just 15 seconds to do over a week’s worth of coding.
Developed by Mphasis Next Labs, the applied research and development wing of Mphasis, the Autocode.AI tool leverages a deep learning framework to quickly prototype solutions.
Jai Ganesh, senior vice-president and head of Mphasis Next Labs, said Mphasis usually conduct “design thinking workshops” with its customers.
“We engage with key stakeholders, from both our customers and internal groups, to come together and ideate and try to solve actual client problems.
“We build intellectual property for the company and deploy them in customer locations, but we also build it for internal consumption within Mphasis,” he said.
“What typically happens is you put a bunch of people together, they go to the whiteboard and start drawing buttons and arrows, creating a login page for a customer, for example, and end up with maybe 50 to 100 wire frames,” he said.
Prior to Autocode.AI, pictures would be taken of the wire frames and sent to the developers, who would then convert these frames into code, which, according to Ganesh: “Could take anywhere from a week to 10 days, depending on the complexity of the scenario being built.”
Read more about AI
- IT leaders share their views on AI and how to take advantage of the technology to benefit their businesses.
- Researchers at IBM are pushing the boundaries of what AI and machine learning can do, but remain wary of the ethical implications that accompany the proliferation of this technology.
- AI has the potential to assist in the analysis of data in a range of topics, but businesses need to understand its limitations and have the right tools to get the most benefit, says a Swiss entrepreneur.
Now, with the introduction of Autocode.AI, this can be done in seconds. “The thought process here is if you’re able to teach the engine enough images, it should be able to understand and generate the appropriate code for it,” said Ganesh.
“We’ve trained the engine on hundreds of thousands of images, and when you put them through the deep learning engine, it breaks them into hundreds of different sub-images. It’s trying understand sub-segments of the image so it can generate the appropriate bootstrap HTML [hypertext markup language] code for it. To go from a raw wire frame to a fully fledged HTML code, it takes around 10 to 15 seconds.”
But Mphasis is not trying to replace its HTML designers. “You still need a human element from a design and aesthetics perspective,” he said. “It’s about reducing the time taken – and, of course, the associated cost, because now the developer doesn’t need to spend time on the nitty gritty, they just need to focus on the aesthetics, the look and feel, while the engine takes care of the rest.”
Autocode.AI also assesses vulnerabilities in the code, automatically fixes bugs, and automates the management and deployment of code in cloud computing environments.
“In the midst of the Fourth Industrial Revolution, enterprises need to put software design and engineering centre stage by utilising advanced deep learning techniques that will bring coding into the future,” said Nitin Rakesh, CEO and executive director at Mphasis.
Ganesh added that Mphasis sees the tool as something to make the lives of its development team easier: “Right now, we don’t plan to commercialise the Autocode solution directly. Things may change in the future at some point but, as of now, we are looking at it as an accelerator – we don’t plan to sell it separately as a licensed product.”