Programming may no longer be required for every digital task, but thinking still is.
How is artificial intelligence changing low-code and no-code platforms? Why will programmers not simply disappear, and what is happening to developer experience (DevX)? This is a direct look at a digital transition in which the work moves from typing every line toward designing, checking and operating whole systems.
Prologue: code at a crossroads
Software development stands between two fast-growing forces. Low-code and no-code platforms promise to reduce the debugging, dependency management and repetitive construction required for ordinary business applications. AI assistants appear able to perform tasks once reserved for junior and even senior developers. Both developments sit within a broader concern for developer experience: the quality, speed and clarity of the entire process of creating a digital product.
This does not necessarily mark the end of coding. It may be the beginning of a wider form of engineering in which judgement, architecture and verification matter more because implementation becomes easier to generate.
Low-code and no-code in practical terms
- Low-code platforms are visual construction environments that allow custom code when a standard component is insufficient. Examples include OutSystems and Mendix.
- No-code platforms aim to let a user assemble pages, data and workflows through visible blocks, controls and scenarios. Examples include Bubble, Webflow and Adalo.
The promise is speed and accessibility. AI adds another layer: instead of serving only as an assistant, it can propose the structure, generate components and help a user navigate the platform.
Artificial intelligence enters DevX
AI is being integrated directly into low-code and no-code environments. In practice it may:
- generate code or a workflow from a natural-language request;
- create and execute tests for application behaviour;
- adapt interface suggestions to observed user needs;
- recommend improvements to logic, integrations and components;
- identify likely performance and security bottlenecks.
Technical building blocks
- Large language models, such as GPT-, PaLM- or Claude-family systems, generate text, scripts and logic.
- Classification and clustering models help analyse user behaviour.
- AutoML supports rapid creation of models based on an organisation’s own data.
- Retrieval-augmented generation (RAG) combines model capability with selected business documents and data.
- CI/CD integration can run automated checks and validation as generated work moves toward release.
The original article cites a 2023 Siemens/Mendix example in which AI-assisted testing, interface improvement and code generation were reported to accelerate delivery. It also refers to a Mendix “State of Low-Code” report and a 2024 Carnegie Mellon study for adoption and productivity figures. Such vendor and survey statistics must be read in their original methodology and should not be treated as universal guarantees. The defensible conclusion is narrower: AI has become part of the platforms themselves, helping people create and improve applications while increasing the importance of expert review.
Examples of AI-enabled platforms
International products
- OutSystems combines application construction with automated testing, code assistance and UX recommendations.
- Mendix applies AI to tests, defect discovery and business analysis.
- Appian focuses on AI-supported automation of business processes.
- Betty Blocks offers a no-code environment with AI integrations for process scenarios.
- Zoho Creator combines generated application elements with machine-learning capabilities.
- Bubble, Webflow and Adalo provide visual application or website construction with growing AI assistance.
Products used in the Russian market
- Craftum is used for site generation and performance-oriented assistance. The article describes a marketing agency automating landing-page creation from structured briefs.
- AppMaster generates server logic and interfaces; the cited logistics example describes a custom CRM delivered much faster than a conventional four-month schedule.
- Tilda adds assistance for content and media selection.
- Moguta.Cloud concentrates on e-commerce experience, traffic analysis and personalisation.
- Nethouse, uCoz and uKit provide accessible packaged tools widely used by small businesses and education.
- Quickbase supports the construction of mobile and desktop business solutions, including industrial scenarios.
The article attributes Russian adoption estimates to TAdviser. As with every market estimate, the date, sample and definition of “using no-code” must be checked before applying the percentage to a decision.
Why developers are still needed
If a platform feels simple, somebody has already solved substantial complexity below the surface. A low-code system rests on a technology stack designed and maintained by qualified engineers. Programmers remain necessary:
- to build and operate the platforms themselves;
- to integrate external systems and data;
- to update, scale and observe production applications;
- to control architecture, identity and security;
- to create unique components where the standard catalogue does not fit the business.
Low-code simplifies work for users precisely because engineers have absorbed complexity into reusable infrastructure.
When visual does not mean simple: BPM and Camunda
Business-process platforms such as Camunda let specialists model processes visually with BPMN and connect those models to executable services. AI can generate a draft model from a description, predict bottlenecks, recommend improvements and check API or workflow consistency.
But a model does not automatically understand the real business meaning. Without an architect, domain expert and validator, a plausible diagram may encode the wrong rule. Visual notation changes the interface to complexity; it does not abolish complexity.
Where low-code fits—and where it does not
| Scenario | Low-code/no-code fit | Conventional engineering |
|---|---|---|
| Prototype or MVP | Usually a strong fit | Not always required initially |
| Corporate CRM/ERP workflow | Often suitable | Needed for complex integration and extension |
| High-load system | Usually limited | Required for control and performance |
| Complex BPMN process | Useful with safeguards | Engineering participation remains important |
| Highly customised product | Suitable only in parts | Usually required |
| Scientific or algorithmic system | Poor fit | Required |
| Marketing landing page | Strong fit | Usually unnecessary |
| Application with exceptional security requirements | Often inappropriate | Required with explicit controls |
Citizen development: the business becomes a builder
No-code and AI have produced the “citizen developer”: an analyst, manager, marketer or other domain specialist who creates a solution without waiting for a full IT delivery team. The potential benefits are clear:
- initiatives can be tested and introduced quickly;
- the central IT queue receives fewer routine requests;
- business knowledge moves closer to implementation.
The risks are equally real:
- shadow IT—solutions created without the security, architecture or support standards of the organisation;
- additional vulnerabilities and excessive permissions;
- poor scalability, duplicated data and workflows with no durable owner.
This increases the value of the developer as mentor and platform steward. An interface may be assembled quickly; understanding the consequences remains team work.
Market direction
The article cites Gartner and Forrester forecasts for rapid growth in citizen development and AI-enabled low-code. Forecast figures should be treated as dated scenarios, not promises. The underlying technology trends are more useful:
- hybrid delivery that combines conventional code with low-code components;
- deeper AI integration into workflows and UI/UX tools;
- greater emphasis on security, governance and scalable runtime foundations.
Conclusion: code is not dying; it is maturing
Low-code, no-code and AI are changing the role of the developer:
- from implementer toward architect and verifier;
- from repetitive manual work toward system design;
- from solitary specialist toward mentor for domain experts and citizen developers.
The strongest future is a combination of simple interfaces, powerful engines and AI used as an assistant rather than an authority. Code will not vanish. It will become less visible in ordinary cases and reappear wherever the problem demands unique behaviour, scale, safety or deep engineering control.
Sources named in the original article
- Mendix, The State of Low-Code;
- Carnegie Mellon research on no-code adoption;
- McKinsey material on low-code development trends;
- Gartner material on citizen development;
- Forrester forecasts for AI-enabled low-code/no-code;
- TAdviser reviews of the Russian low-code/no-code market.
