No-code, low-code,
what code?

Just like almost everything in life, coding is on a journey. In fact, how code is written is evolving. More and more, it’s passing through ‘no-code’ or ‘low-code’ platforms on its journey towards AI-assisted development. Are you ready to learn more?

What do you mean by ‘low-code’ or ‘no-code’ development?

The terminology can be misleading. This is because the distinction isn’t necessarily about how much code is required to solve a problem, but rather who is involved and the scope of the problem in question.

On the low-code side, you’ll find professional developers, expert system integrators, and enterprise architects, all with extensive technical abilities. Low-code is all about re-using, streamlining and simplifying integration tasks to achieve an enterprise-grade integration scenario or application. Low-code means more than building apps without code, as the platform Mendix talks about.

On the ‘no-code’ side, business users aim to solve a relatively simple integration problem – often a ‘point-to-point’ or trivial business process one – but without needing any software development skills or to even write a single line of code. This is the so-called citizen-development scenario.

Is there a business or practical use to that though?

Generally speaking, we can say no-code/low-code development platforms allow you to create software using GUIs (graphical user interfaces) instead of the standard computer programming and common coding techniques. There are so many examples geared to building apps for enterprises or small businesses. Have a look at giant solutions like Salesforce Lighting Platform, Microsoft Power Apps and Google App Maker. Or the underdog competitors such as Zoho Creator, Appian and Kony Quantum.

Go on...

Even better, if you’re looking for a solution to build a mobile application, there’s a long list of no-code solutions from the likes of AppSheet, BettyBlocks, Dropsource and QuickBase. Ready to build your first app?

I need some more help…

A new trend in designing websites – or other UI based systems – is ‘Designing in Code’. This is where instead of making non-interactive mock-ups, designers go straight to the code to make quick and easily changeable fully-functioning products. This approach also draws on techniques outlined in the book Atomic Design by Brad Frost.

However, the reality is that not all designers are coders.

This is where AI-assisted development comes in, translating part of what the designer does into ‘code’. In October of 2018, Gartner published its ‘Top 10 Strategic Technology Trends for 2019’. Third on this list was AI-driven development, and its ability to equip non-developer professionals with AI-driven tools to generate new solutions automatically.

You mean a computer programming by itself?

Yes. You can already see simple versions of this with website-making services such as Wix, or the awesome Sketch2Code example. Usually, several programmers draft their ideas or mock-ups on paper, then proceed to code. But imagine if your ideas could be transposed from that piece of paper to thousands of lines of code instantly? Services such as the already mentioned Mendix are taking low-code platforms and elevating them with AI-assisted development. The belief that ‘every company needs to become a software company’ has led to software that empowers business experts, while also empowering professional developers through abstraction, automation, and intelligent assisted development.

AI can already help coders perform many different tasks, such as code analysis, bug fixing, code optimization, code hinting and, as with Sketch2Code, code generation. AI can learn a coding ‘language’ and become an assistant to a developer. For example, Mozilla and Ubisoft partnered to make an AI assistant called Clever-Commit specifically to help with coding.

So AI in this case is helping non-developers develop, as well as speeding up the job for professional developers in various ways. This includes debugging – the AI uses a deep learning algorithm to flag errors and speed up the debugging process. Or, in the case of, TPOT – Data Science Assistant and Google AutoML, to speed up Machine Learning algorithm training. Do you know Google AutoML has already taken part in a Kaggle competition?

So, are we moving towards computers programming computers?

General purpose coding is still very much a human-only skill, while generated code is only possible for very simple or mechanical tasks. However, AI has been very good at automating code-related tasks, such as review, bug fix and optimization. Auto ML is not the ultimate solution for any Machine Learning problem, but a time-efficient way to speed up prototypes and demo development.

Stay in touch Stay in touch

Receive R20 via e-mail!

Is not Authenticated