News

Mendix Improves AI Engine for Low-Code Development Assistance

• By David Ramel
• 04/19/2019

In announcing the Spring '19 release of its low-code/no-code software development tooling, Mendix touted the second generation of its artificial intelligence engine to help with code completion suggestions and assistance.

That AI engine powers Mendix Assist, which can provide suggestions for what developers are likely going to type as they work in the company's Mendix Studio (no-code) and Studio Pro (low-code) developer environments.

Putting AI to use as a virtual assistant for computer programmers in low-code and professional IDEs may signal a growing trend, according to industry analysts and vendors such as Mendix.

The company said Mendix Assist can help non-programmers or newbies -- more likely to use Mendix Studio -- with AI-assisted next-step-suggestions that can serve to teach them how to build apps. "Mendix Assist is the equivalent of having a world-class coach looking over their shoulder to dramatically accelerate the learning curve without the added cost of additional training resources," the company said last June when Mendix Assist was unveiled.

For more skilled developers who might be using the low-code Mendix Studio Pro, the company said the AI engine can increase productivity and capacity as well as serve as a virtual assistant for collaboration. "Mendix Assist's AI-enabled pair programming helps experienced coders by providing high-accuracy suggestions for the next modeling activity, reducing coding mistakes to facilitate expert quality software," the company said this week. "The productivity gains and enhanced speed enabled by AI suggestions frees developers to work on the creative, impactful side of application development to solve critical business needs."

Putting AI to use in pair programming may become a growing trend, as Mendix pointed to Gartner Inc. research predicting "by 2022, at least 40 percent of all new AD (application development) projects will use virtual AI co-developers on the team."

Much like Microsoft using machine learning models based on quality development projects to improve its IntelliCode functionality in its Visual Studio IDE and the Visual Studio Code editor, Mendix said its Mendix Assist AI recommendations are based on 10 years of software development experience in the Mendix development community, resulting in a data set of some 5 million anonymous application logic flows gathered across 15 industries. "Mendix's AI algorithm learns with continued use, with the result that its recommendations have achieved 95 percent proven accuracy," the company said. Previously it had claimed 90 percent accuracy, pointing to the second-generation boost.

"We're finding that AI-enabled development is driving innovation throughout the enterprise," said Johan den Haan, CTO at Mendix. "Ideas can come from all corners of the organization. AI-assisted development enables a broader group of people to develop applications. The AI assistant nudges them in the right direction, making them more productive as well. Increasingly, visual models and AI-assisted development are becoming key ingredients for successful, software-driven companies."