The pandemic has driven many tech trends, including a shift further towards a mobile-first world.
More people are turning to apps than ever before for essential services like banking, food delivery, connecting with distant family, or just finding entertainment during these times of social distancing.
As Frank Moyer, CTO of Atlanta-based Kobiton, told Hypepotamus, the explosion of mobile-first customers is putting additional strains on testers and developers, whether they are working on an app for a Fortune 500 company or a startup.
“The mobile user is a very demanding user…[much more than] someone using a desktop. A mobile user will abandon an app if they consistently get a response time greater than two seconds,” Moyer said.
That’s a lot of pressure for a team trying to rollout or update an app.
To add to that pressure, each device type and model has different requirements, meaning that testing on a device like an iPhone 8 won’t determine if the app will necessarily work on an earlier phone model.
When you consider all the different phones and tablets on the market from the likes of Apple, Google, and Samsung, testing one app across all available devices can quickly become unscalable.
Currently, Moyer says that it takes an automation group about two weeks to update their scripts. “Right now, the testers are in a world of hurt when it comes to resources and timelines,” added Moyer. “They are often pushed against aggressive deadlines to complete testing…and when you add on to that the multiple mobile device types — and the many manufacturers and operating versions — it’s very different from the web. It’s a somewhat impossible task.”
The Beauty of AI in mobile testing
Kobiton’s team thinks AI can help tackle the seemingly endless task of mobile testing.
Kobiton’s platform uses statistical techniques and deep learning to augment what a company’s testers and developers are working on. With machine learning and visual analysis techniques, Kobiton can identify visual and performance issues and how they will show up on different devices.
This can be anything from skewed font heights to text alignment problems; design elements that can turn mobile users off an app quickly.
Beyond just looking at the functionality, Moyer also told Hypepotamus that it is important to quantify what makes an app “beautiful.” This means training Kobiton’s AI to analyze the components and design principles of top 50 mobile apps like Amazon and Uber. Those findings are then compared against a client’s app to see what design elements need to be re-worked.
Moyer has been working in AI and machine learning throughout his career. He was approached by Kobiton’s CEO to fill the CTO slot and ultimately helped bring AI to the mobile testing platform.
For Moyer, Kobiton can help companies add continuous testing to CI/CD (continuous integration and continuous deployment).
To date, Crunchbase reports that Kobiton has raised just over $22 million, including a Series A led by BIP Capital.
In November of 2020, the team acquired an Atlanta-based on-premises mobile testing platform, Mobile Labs Inc.
Feature image from @hckmstrrahul
