Will GenAI Spell the Death Knell for Low-Code/No-Code Platforms? Quite the Contrary

The CEO of Stability AI, Emad Mostaque, stirred the internet pot sometime recently when he declared that there will be no human programmers in five years. He seems to have stretched the canvas further with a more aggressive and apocalyptic prediction, this time training his guns on outsourced coders in India, who, he said, would lose their jobs in two years! Well, this prompted many internet and social media gatekeepers to cast aspersions on the “stability” of this assertion, but then his company is called “Stability AI”, and people did not expect this “un-stable diffusion” of information from him. We need to be kinder to the ones who believe they can change the world, as Steve Jobs reminded us in his famous take on these “square pegs with round holes”.

But truth be told, much before GenAI came into existence, scientists and engineers have been working hard on creating low-code/no-code platforms, especially in the field of Data Science and Machine Learning. One has to only look at the rapid ascent of these platforms in Gartner’s Magic Quadrant for Data Science, from 2017 onwards, to see that these low-coding platforms have been “eating the world (err…Quadrant)”. In fact, low-coding platforms like Dataiku, Alteryx, KNIME, and Rapid Miner have been completely dominating the top quadrant over the last five years.

Now as we all know, the brilliant Simon Sinek taught us that to inspire action, we need to start with the “why”. Similar to how great leaders inspire action by starting with the why, so do some great innovations, which have fundamentally changed and shaped the way of human life and existence. So, let’s get started.

Why do we need Low-Coding/No-Coding platforms?

The why has three components to it: 1) Democratization 2) Industrialization 3) Efficiency

#1. Democratization – We can articulate it as “AI/ML and data science for many”. Data science provides organizations with differentiated competitive advantages, but there remains a dearth of readily available data science talent. Many progressive organizations have therefore opened data science and AI/ML to nontraditional roles, such as the “Citizen Data Scientist”. Gartner coined the term Citizen Data Scientist to classify people whose primary function is outside of the field of statistics and analytics, but are leveraged by an organization with some amount of training to do predictive and prescriptive analytics. The low-coding/no-coding platforms are an excellent tool that enables the proliferation, application, and scaling of AI/ML and data science across the business.

#2. Industrialization – We can articulate it as the “Rapid scaling of data science and AI/ML” across the business. The “click and drag” as well as the “plug and play” features of these platforms, using reusable components (read “accelerators”), help in the creation of new solutions at roughly half of the time it takes using traditional coding tech stacks. (SAS/R/Python + Analytics Database (Relational/Columnar)). For example, in the “coding world”, if we are able to build a few models (single digits), in the “low-coding world”, we can build thousands of them in the same time frame.

#3. Efficiency – Let’s articulate this as the development of data science solutions “With half the people, in half the time”. If you needed 20 people to do 20 Data Science Tasks (all kinds of descriptive, prescriptive, predictive analytics) in 20 days in the “coding world”, you will need 10 people to do those same 20 Data Science tasks in 10 days in the “low-coding world”.

How are Low-Coding Platforms changing the way we work? How is working on a typical data science project different?

Teams that are using low-coding platforms are now able to do analytics in real-time and on the fly during meetings. This is huge and transformational. The whole paradigm of waiting for weeks and months for business questions to be answered goes out of the window. Ad hoc requests can be answered in real-time, almost immediately, as the request forms in the mind.

In the “coding world”, analysts have to query databases to fetch the data (Analysts need knowledge of some form of SQL) and then do analysis on that data using SAS/R/Python and then present the results in a visual form (some BI tool like PowerBI, Tableau, Qlik, MSTR, etc.). In the “low-coding world”, you can do all of this on the same platform, and you don’t need to staff the team with three different kinds of skill sets. Additionally, in the “low-coding world”, one also has the flexibility to write codes in R/Python and insert them into the native workflows of the platform, thus making it a haven for “coders” as well, who can now spend more time on the analysis, insight generation, model accuracy rather than on trying to get the syntax right, flipping from one platform to another. This represents a completely new way of working on Data Science projects.

What are the benefits of using a “Low-Coding Platform”?

The answer is twofold – “top line impact” as well as “bottom line impact”. More decisions can be taken in a data-driven manner, and more AI/ML-driven intelligent decision-making systems can be integrated into the business as these platforms enable democratization and industrialization. This has a direct impact on the top line of the business. Additionally, companies can reduce their data science teams by half and still get the same amount of work done using these platforms. This has a direct impact on the bottom line in terms of dollars saved, giving companies economies of scale.

Lastly, coming back to the fundamental question of “whether GenAI will herald the end of the low-coding/no-coding platforms”. Well, in my view, the answer is an overwhelming “no”. Even if GenAI is used to generate code, the knowledge of how coding works to build Analytics and AI/ML solutions will still be crucial in using code generated by LLMs. The codes written or proposed by LLMs will still be cryptic and not readable by the teeming majority of non-coders. It will not enable the citizen data scientists to understand the code and deploy the data science solutions. This is where the low-coding platforms carry the edge. The visual coding schema of the low-coding platforms has a level of simplicity and transparency that cannot be matched by anything that has been invented till now. Maybe, someday the LLMs will be able to produce the workflows built by these platforms, similar to how they generate code in coding languages today. The end result would still be the same. The world will move towards the universal adoption of “low-coding platforms”.