Update (June 2026): This piece started life in April 2022. I’ve revisited it twice since — once for 2023, and now in 2026. The 2026 update is the biggest of the three: low-code is no longer the contested concept it was four years ago, so several of the original myths are barely myths anymore. I’ve kept the ones that still come up in client conversations, retired the ones that don’t, and added the new myths that have grown out of the AI conversation — questions like “won’t ChatGPT just replace low-code?” that nobody was asking in 2022.
Four years ago, when I first wrote this article, I opened it with a Gartner prediction: by 2024, 65% of application development would happen on low-code or no-code platforms. We’re past that date now, and depending on whose numbers you trust, we either landed at that figure, overshot it, or are still a year or two short. What’s not in doubt is the direction. The 2025 Gartner Magic Quadrant for Enterprise Low-Code Application Platforms — published July 28, 2025 — describes a market where six platforms (Mendix, OutSystems, Microsoft Power Apps, ServiceNow, Appian, Salesforce) sit in the Leaders quadrant, and where “AI-assisted tooling, composable architectures and built-in governance” are now baseline expectations, not differentiators.
So the conversation has shifted. Four years ago, people asked whether low-code was real. Today they ask whether it still makes sense in an age of agentic AI coding assistants. Both questions have answers worth talking about. Below are the myths still worth addressing in 2026 — some unchanged from 2022, some retired, and a few new ones that the AI era has created.
Many companies start their adventure with low-code by looking for the “perfect” platform. If you’re here to learn which is the best tool out there, I have some bad news for you – there’s no such thing. Sure, there are some market leaders, and some solutions – like Mendix, OutSystems, or Microsoft Power Apps – are more popular than others. There are also some that are considered challengers – Oracle APEX is a great example here.
But the reality is that low-code technologies have widely different use cases, and a platform that’s great for one thing, might not be all that good for another. There is also no point in trying to implement a platform that doesn’t sync well with your company’s technological stack, just because you saw it in the Gartner’s Magic Quadrant. You’ll just make things harder for yourself. Research is the answer here – ask your in-house team to get the necessary information (alternatively, if you don’t employe IT specialists, look for low-code consultants from outside the company).
Let’s get one thing straight: sure, low-code is usually substantially cheaper than traditional software development. However, that doesn’t mean it’s always incredibly cheap, and there are some circumstances when it can actually get quite costly.
One of the most important things to take note of here is the licensing model that the company behind a given low-code technology uses. Some of these tools are licensed in complicated ways, with the final price being dependent on such things as the number of active users, developers, etc. This can easily make costs grow if you’re not careful, so you should always make sure to consider and evaluate the subscription plans and models before you decide on a specific tool. And don’t just look at what it means for you right now – try to also take the future into account.
A reminder of how this can bite: Microsoft Power Platform’s per-app and per-user licensing models have changed multiple times since 2022, and what looked cheap in a pilot can look much less cheap once you scale to 500 users across three apps. We’ve seen clients restart vendor selection from scratch over exactly this.
It’s one of the most common untrue things I hear about low-code technologies: that you can’t create complex projects with such tools. This idea is completely made up. I mean, sure, low-code is great for simple applications you can make in a day or two, but it can also be used to create powerful, flexible solutions for big enterprises (a CRM or ERP system, a freight management platform, a warehouse organization system etc.). Pretius is a great example here – we’ve successfully used technologies such as Oracle APEX to create full-blown enterprise-grade systems for big, global companies (I’ve talked about it during a webinar organized by Oracle, so check it out, if you’re interested).
Low-code can be used to create anything – assuming you have the necessary knowledge.
Low-code is often considered to be a solution made specifically for people outside of the IT world – business users who can learn to create apps, becoming so-called “citizen developers”. While it’s certainly true that these platforms can be used by people without coding knowledge and/or experience, it by no way means they are only meant for such users. Software developers are still an important part of the process – their presence makes some things much easier, and significantly extends the list of available possibilities.
For many companies, low-code is something of a shortcut – one that allows them to make great solutions faster and cheaper thanks to the use of various drag-and-drop “wizards” and read-made elements. It also standardizes the software development process, which can be a huge boon.
However, it’s important to remember that all of this is still based on programming languages, databases, and so on – it’s made a bit easier to use, with less coding required, but you’re not limited to these read-made elements and functionalities. You can extend the platform and software made with it in various ways, though it will probably require more knowledge and actual software development experience.
Another myth that I sometimes encounter is that low-code platforms can’t be used to create full-blown software products that you’ll later sell to your clients. I can actually prove this isn’t true by showing you an example – Pretius has a proprietary sales commission system (MonkeyJar), which is a product created with the Oracle APEX low-code platform.
In fact, in some circumstances – for example, when you’re running a startup with a rather limited budget, and an aim to create a product with the best possible time to market and validation time – low-code can be a truly great choice. In such scenarios, it’s a much, much better and safer alternative to traditional software development.
Some people also believe that low-code software is hard to maintain but the reality is that it just works differently than in the case of “traditional” – say Java-based – systems. You have to change your mindset. The main reason for that is the fact that the apps usually exist on the platforms they were made with, and are operated through them.
In fact, in a way, low-code maintenance can even be a little bit easier than what software developers are used to because the platform’s creators provide users with a clear methodology – a path to follow, an explanation of the entire maintenance process, and the application’s life cycle. Instead of coming up with your own maintenance strategy, you simply have to follow these guidelines.
Another common myth is the belief that applications created with low-code technologies can’t be modified in any way, and can’t be made to work with other tools. This is very untrue, as most platforms offer extensive integration capabilities. Low-code apps and systems can easily be used with various components from other vendors.
However, once again, be sure to check the information about integrations on the website of the platform you’re interested in using – in most cases, there’s detailed documentation available. If something important for your project isn’t on the list, consider choosing a different low-code solution, or get ready for additional expenses.
Now, this is kinda true, at least on the surface. If you look on a portal such as LinkedIn, for example, the number of specialists who list low-code skills and experience on their profiles isn’t very high, to say the least. It’s important to know, however, that this is only a part of a much bigger picture.
The key to understanding the situation is something I’ve already mentioned in this article – low-code platforms are based on existing technologies and solutions. This means that people familiar with them can easily become low-code developers, even if they don’t have experience with low-code itself. Learning these tools is quite a simple matter, especially for someone with extensive IT knowledge, and after a short onboarding process, such developers will be able to work with low-code tools without too many issues. For example, someone who knows Oracle DB, SQL and PL/SQL will have no problems with an Oracle APEX-based project.
So, while the number of specialized low-code developers might not be very high, the number of potential low-code developers is.
This is the question I get most often in client meetings in 2026, and it’s the one with the most interesting answer.
The short version: no, but the long version has nuance. Generative AI assistants are extraordinary at producing code. They are also extraordinary at producing plausible-looking but quietly wrong code, especially in domains they haven’t seen much of (enterprise database schemas, niche compliance constraints, integration with 1990s middleware). Low-code platforms aren’t competing with AI assistants — they’re using them. APEX 26.1 ships with APEXlang, designed specifically to give AI coding agents a supported, governed way to generate and modify APEX applications. Mendix and OutSystems have similar AI-assisted authoring flows. The future isn’t “AI replaces low-code.” It’s “AI generates inside low-code, with the platform providing the guardrails.”
Forrester’s recent research on this — they call the AI-coding-agent layer “TuringBots” — concludes the same thing we see with our clients: AI dramatically accelerates low-code adoption, particularly for citizen developers. It does not displace it. Natural-language prompts are an incredible way to start an application. They are a terrible way to maintain one.
“Vibe coding”, which is a clever term for building software by chatting with an AI agent until something works, is a real and useful workflow for prototyping and personal projects. We use it ourselves for throwaway scripts and proofs of concept. But it’s not how enterprise software gets built and maintained, and it won’t be in 2026, 2027, or any year we can credibly forecast.
The reason is boring: enterprise applications need governance, audit trails, role-based security, deployment pipelines, and disaster-recovery stories. Vibe coding produces none of these on its own. Low-code platforms produce all of them by default. When a regulated client asks “Where is the access log for who modified this approval workflow”, “I asked Claude, and it gave me this code” is not an acceptable answer.
The “we’re already a Microsoft shop, so Power Platform is free” argument is one of the most common starting positions we hear, and it’s also one of the most expensive misconceptions in the room.
Power Platform’s pricing structure is genuinely complex. The capabilities that come “free” with Microsoft 365 are limited (basic Power Apps for standard connectors, Power Automate for Office 365 connectors only). Anything substantial — premium connectors to external systems, AI Builder, custom APIs, Dataverse beyond a certain size — moves you onto per-user or per-app plans that scale very quickly. For a 500-user enterprise app with a few external integrations, the licensing arithmetic often ends up higher than equivalent deployments on platforms with simpler pricing models (Oracle APEX is the obvious example — it’s included with the database license you already pay for).
None of this means Power Platform is bad. It means “we already have it” is not a sufficient reason to choose it. Run the math.
Looking at this list four years after I first wrote it, the most interesting thing isn’t which myths are still around. It’s which ones have quietly disappeared. Nobody seriously asks anymore whether low-code is “real” or whether you can build enterprise systems with it — the answer to both is now obvious. What’s replaced those conversations is a set of newer questions about AI, agentic workflows, vendor lock-in, and how to choose between five Leaders that all look the same on a Gartner chart.
If you’re working through any of these — whether it’s an old myth from 2022 or a new one from 2026 — drop us a line at hello@pretius.com. We’ve been doing low-code for nearly a decade and we genuinely enjoy these conversations. The initial consultation is free.
