How Mendix Supports AI-Augmented Development

What features does Mendix offer for AI-Assisted Development (AIAD)?

With Mendix Studio Pro, we have been busy setting the foundations to help you use Generative AI for software development.

Mendix chatbot: MendixChat

Since the launch of Mendix 10, we’ve released MendixChat. This is a smart chatbot that answers any development question you may have while working in Studio Pro.

MendixChat is based on a large learning model that we trained on our own data. Want to know how to configure push notifications? Simply ask MendixChat, and you’ll get the answer without having to scour through documentation and forums.

We see MendixChat benefitting new users of the platform to get up to speed more quickly. The faster you’re able to understand how to use Mendix, the faster you’re able to bring value-driving software to market.

When it comes to AI and Mendix, MendixChat is a foundational feature. Our goal is to not just have a helpful chatbot that can answer any development questions and provide real-time guidance. That’s great. But we eventually want the platform to automate tasks and facilitate app generation.

Stay tuned for more, but the prospects are exciting and MendixChat is just the first step.

Logic Bot – new and improved

Mendix has already incorporated AI to accelerate development, particularly with the Logic Bot. Since the release of Mendix 10, this feature has undergone significant enhancements and is now GA.

As you build your applications in Studio Pro, Mendix’s Logic Bot provides predictive, logical next steps to take. With this latest version of Logic Bot, Mendix customers will notice that it’s different looking. We’ve updated the UI to make it easier to use.

The second thing you may notice is that the predictive AI is smarter. It’s more accurate. Working on a microflow? Type what you think the parameters are and it’ll predict them.

Logic Bot is contextual, so it knows what you’re working on and is smart enough to provide accurate real-time guidance.

Build AI-augmented applications with low-code

We understand that AI is critical for your business, too. We haven’t only enhanced our platform with AI to augment developer productivity. We’ve also been busy devising ways to help you deliver AI-augmented applications.

We want you to build smart software that moves the needle in terms of business value.

AWS connectors to AI services

Looking to add image recognition to your software? Perhaps text translation? Or text-to-speech?

There are many “AI as a Service” (AIaaS) APIs out there that you can use to do that without the need to develop, maintain, or directly manage the underlying infrastructure.

Creating those AI-augmented applications requires integrating AI services into your software. That can be time-consuming. But Mendix provides the most flexible architecture for AI-enabled software, offering connectors to AWS services like:

• Bedrock: a fully managed service that makes high-performing foundation models (FMs) from leading AI startups and Amazon available for use through a unified API
• SageMaker: build, train, and deploy machine learning models
• Textract: extract textual content, forms, and tables from scanned documents
• Rekognition: analyze images/videos to detect objects, faces, and content, as well as to perform real-time facial recognition and identification
• Translate: real-time and batch text translation across multiple languages
• Polly: convert text into lifelike speech to create spoken content in multiple languages and a variety of voice character

Machine Learning Kit

The beauty of the aforementioned AIaaS is that they’re there, ready for you to use. But, we recognize the drawbacks too: cost, latency, and data privacy. Our Machine Learning Kit helps mitigate these issues.

Using AIaaS can get very expensive really fast. As an example, a talented Mendix Evangelist created an AI-augmented application demo for internal purposes and he ended up accidentally racking up a $10,000 bill in a matter of days.

Mendix’s Machine Learning Kit allows developers to deploy a machine learning model—built using a common machine learning framework and language—directly within your Mendix application’s runtime. This reduces the latency associated with calling hosted services and allows you to keep private or sensitive data within your own environment.

PostNL, the Netherlands’ largest parcel delivery service, has been working with machine learning for many years to help manage the 1.5M delivery requests their organization processes daily. With the Machine Learning Kit, PostNL has been able to achieve sub-millisecond response times from their model and does not have to share customer data with a third-party, hosted service.

If you’re looking to save costs and provide software that moves faster, then Mendix’s Machine Learning Kit is the feature you need. It is GA.

Just the beginning of AI and low-code

Just like everyone else on the planet, we’re excited about the prospects of AI. We also understand the inherent risks that come with it and how complicated it is to use it well. It’s a big boulder for everyone.

What I’ve discussed here is just the beginning. As Mendix 10 continues to grow, so too will Mendix’s AI capabilities. All in service of helping you deliver the software your users need, faster.