IT Enablers Global Contact Us

Home / Low-Code or Custom Development in the AI Era: Revisiting Our Advice Two Years On

Low-Code or Custom Development in the AI Era: Revisiting Our Advice Two Years On

Blog · September 23, 2026

In June 2024 we published Low-Code or Traditional Development: What shall we go for?. Our advice then was simple: low-code for quick wins, prototypes and business process automation, and traditional development for complex, performance-critical applications that need full control.

Two years is a long time in this industry. AI has changed both sides of that comparison, and not in the same way. So we are revisiting our own advice.

What we said in 2024

Low-code platforms were fast, affordable and easy to use, with built-in connectors. They suited prototypes, internal tools, dashboards and workflow automation.

Custom development, for example on .NET or open-source stacks, gave you control, scalability and the freedom to meet unusual requirements. It suited custom products, real-time and high-performance systems, and complex business logic.

That was a fair picture at the time. Here is what has moved.

Low-code has grown up

The biggest change is that AI now works across the whole life of a low-code application, not just the first draft. Zoho Creator, the low-code platform we implement most often, shows how far it has come.

Two further things make Zoho Creator stand out. Every application is automatically available as a native iOS and Android app, so field teams get mobile access without a separate build. And Creator sits in the same suite as Zoho CRM, Books, People and Desk, so an operational application shares customers, invoices and employees with the rest of the organization instead of copying them, with Zoho's regional data centers keeping that data in the region you choose.

Two years ago, "low-code" often meant "fine for a prototype." Today it is a serious option for core operational systems.

Custom code got cheaper too

The other side has moved just as much. AI coding assistants have made custom development faster and less expensive. A small team can now produce in weeks what used to take months, and building a prototype to test an idea costs a fraction of what it did.

That matters, because cost was one of the strongest arguments for low-code. The gap has narrowed from both directions.

What has not changed

Some things are still true, and it is worth being honest about them.

Custom code is still more flexible. If you need real-time performance at scale, an unusual user experience, deep integration with specialist systems, or a product you will sell and want to own outright, custom development still gives you control that a platform cannot.

Low-code still has platform limits. AI builders are impressive on structure, but complex logic still needs an experienced developer. In our own testing, a platform's AI assistant built forms and reports easily but declined to write a multi-step stock control workflow, and recommended that a developer write it by hand. Self-hosted editions of low-code platforms can also offer fewer features than the cloud version.

Licensing works differently. Low-code platforms usually charge per user. For a large user base, a custom application with no per-user fees can cost less over five years, even if it costs more to build.

AI does not replace judgment. Whichever route you take, someone still has to understand the business process, design the data model, decide what not to build, and keep the system secure and maintained.

How the picture looks in 2026

20242026
Low-code build speedFastMuch faster, with AI agents from requirements to app
Low-code complexity ceilingSimple to moderate processesMost operational systems, with a developer for complex logic
Custom build costHighNoticeably lower with AI-assisted development
Custom flexibilityHighestStill highest
Hosting choiceLow-code tied to vendor cloudLow-code can run on major clouds or on premise, with trade-offs
Ongoing costLow-code subscription vs custom maintenanceSame, and per-user licensing still matters at scale

Our updated advice

We now start with three questions rather than one:

  1. Is this a business process built on data you already hold? Sales, inventory, service, approvals, compliance tracking. If yes, AI-assisted low-code is usually the fastest and most cost-effective route, especially if you already use a suite such as Zoho.
  2. Is this a product, or does it need performance, scale or an experience a platform cannot offer? Customer-facing apps, real-time systems, marketplaces, anything you plan to sell. Custom development, now faster and cheaper with AI, is the better fit.
  3. Does it need both? More and more often, the answer is a low-code core for operations with custom components where they add real value, such as a mobile app for field teams or a specialist integration.

In 2024 the question was "which one?". In 2026 it is more often "which part goes where?".

How we can help

IT Enablers Global works on both sides of this decision. We are a Zoho Premium Partner delivering low-code and Zoho implementations, and we run an AI-led product development practice for custom applications and mobility solutions. That means we can recommend the right mix for your situation rather than the one approach we happen to sell.

If you are weighing up a new system, talk to us.