The Citizen Developer

Where Did Low Code/No Code Go?

Don't look now, but the era of low-code/no-code development may be passing us by faster than we ever dreamt it would.

• By Howard M. Cohen
• June 26, 2024

It seems like only yesterday we were marveling at the applications we could develop by simply moving icons around the screen to assemble processes. Excitement grew as more and more new low-code/no-code (LC/NC) platforms were introduced to help us do that better and better. Soon we found ourselves with an abundance of choices, perhaps even an overload.

But if you blink your eyes too long, you might miss that entire era. It already looks as though icon-based application building may be becoming a thing of the past.

We May Be Spared – But Not LC/NC

The brightest minds on the planet, including Bill Gates, Elon Musk, and the late Stephen Hawkings, have warned of the potentially devastating existential consequences generative artificial intelligence (GenAI) could bring. What happens when the machines realize that the greatest threats to the planet are human beings?

Whatever you think about those warnings, it now seems inevitable that we're eventually going to stop using LC/NC, preferring instead to simply tell our AI what we need our applications to accomplish.

This transition will surely start out innocently, perhaps we could even say sneakily. For example, in a recent blog post about how Microsoft's own citizen developers get the most out of the Power Platform, the company innocently explained that "We're infusing Microsoft Copilot into the platform so you can use AI to convert your natural language queries into solutions."

Copilot has perhaps already been the biggest of the big bets Microsoft has ever made on a product launch. The company pointed out that, although the mobile phone took 16 years to get from introduction to 100 million users, and the internet took 7 years to do the same, GenAI took only 3 months.

For companies like Microsoft, that's just too irresistible an adoption rate to ignore.

But folks in Redmond are not alone! In my March post, I reported on the advent of TuringBots, which "use AI and machine learning (ML) to build models that 'learn' from existing code and identify which code generator can meet the business applications and infrastructure requirements to generate and deliver source and executable code." Other developers busily incorporating GenAI into their LC/NC platforms include ChurchAPT, IDP, Amazon Q Developer, DataRobot, Dataiku, Google Cloud AutoML, Airtable AI, Appian, and even industry leader Mendix.

A Gradual Transition

What I'm predicting is a somewhat gradual transition from coding to LC/NC to AI-Augmented LC/NC and finally to "Just Ask AI!"

Already, developers are explaining how AI helps citizen developers build intelligent applications that simplify complex tasks and execute repetitive functions. They see that how the integration of AI into the development process is improving the efficiency with which new applications are created, tested, and rolled out, and vast improvements in error-checking and correcting. And they see how the technology lowers the barrier to entry into citizen development, enabling faster innovation, reduced costs for application development, increased agility, better security, enhanced information governance, and more.

If all this has the familiar tone of that infomercial pitch for Ginsu knives (They slice! They dice! They chop. They mince. They perform minor surgery!), that's because these are the same claims we heard for all the LC/NC platforms, and even the coding development languages we started with. Just better, faster, and less expensive. (And if you act right now…!)

The folks at NextWorld suggest four ways in which non-programmers (citizen developers) should think about AI today: first as a personal assistant; then as a "recipe follower" cleverly suggesting that "every algorithm is a recipe;" than as a detective, which sounds a bit like pivot tables on steroids; and finally they suggest that "an AI-powered citizen developer is a gardener of the algorithm."

All of that sounds nice, gentle, and pleasant, but let's be honest with ourselves about what we're ultimately looking for: computers that simply obey spoken orders to perform all manner of functions, kind of like the Star Trek computer.

Likely Outcomes

It seems most likely to me that eventually citizen developers will be called upon to produce the logistic equivalent of a functional specification describing all the functions, relationships, processes, and other actions an application will be required to perform. This will be submitted to the GenAI which may discuss it with them, ask questions, seek clarifications, and ultimately code the necessary software to fulfill that functional spec.

While this may sound more involved than the "natural language programming" promised by so many, the likeliest outcome will be, like everything else, a hybrid. Quick routines will be available for the asking, with more complex software requiring a functional spec.

If that sounds like it fully replaces the programming functions of your IT department, that's only because it does.