Last CallLast Call Pool

No-code

Najczęstsze błędy w tworzeniu aplikacji no-code

September 20, 2024 3:15 PM

Long Lecture

🇵🇱 Polish

Bratysława 1

About

Codziennie widzę, jak początkujący programiści no-code próbują tworzyć aplikacje i ponoszą porażkę. Omówię najczęstsze błędy popełniane podczas tworzenia aplikacji na platformach no-code, takie jak nieodpowiednie planowanie, nadmierne komplikowanie projektów i nieefektywne mapowanie procesów. Przedstawię również najlepsze praktyki, które pomagają uniknąć tych problemów, poparte rzeczywistymi przykładami.

‍

Talk by

Photo of Marcin Kowalski, CEO at Archman

Marcin Kowalski

Watch the full talk

Watch this WaysConf session, then continue with related talks or explore the current programme.

Explore the 2026 agenda More WaysConf recordings

From the recording

Talk in brief

Marcin Kowalski, CEO of Archman, discusses the practical challenges and common mistakes of building large-scale applications using no-code platforms. Drawing from 17 years of experience and the evolution of their platform Navigator 365, Kowalski explains how no-code has transitioned from rigid system blocks to highly customizable generic objects. While no-code excels at managing dynamic workflows, user permissions, and dashboards, it faces severe limitations in advanced logic, complex integrations, and version control. Kowalski compares no-code with traditional coding, highlighting how generative AI has revolutionized coding by making it accessible to non-programmers, thereby shifting the balance back toward low-code or hybrid approaches. He recommends a balanced strategy: using no-code for volatile business processes and user interfaces, while relying on universal programming languages and custom corporate APIs for heavy integrations and complex backend calculations.

Key takeaways

01
Evolution to Generic Objects
Modern no-code platforms succeeded by moving away from rigid, pre-defined system fields like counterparty or goods in favor of generic, fully customizable objects. This shift allows creators to define any object's properties and forms from scratch.
Watch from 4:20
02
The Integration Bottleneck
While no-code handles simple JSON and REST API calls well, large enterprise projects often require dealing with legacy XML, database views, and flat files. These complex integrations are far more efficiently managed through traditional code rather than visual no-code blocks.
Watch from 11:16
03
Generative AI Empowers Traditional Coding
Generative AI has transformed software development by easily writing connectors and debugging code, making traditional programming accessible even to non-technical users who can understand and tweak the AI's output.
Watch from 18:41
04
Strategic Division of Labor
To build successful large-scale applications, organizations should use no-code for frequently changing elements like workflows, dashboards, and permissions, while coding advanced logic and heavy integrations.
Watch from 20:42
05
The Version Control Challenge
Managing version control and merging updates in large no-code applications remains a major obstacle because visual tiles and blocks cannot be easily merged or compared like lines of text.
Watch from 26:04

Video chapters

0:31Introduction and Company Background
Marcin Kowalski introduces Archman, their platform Navigator 365, and how the company was founded by non-IT professionals using early low-code tools.
2:15The Evolution of No-Code Architecture
The speaker outlines their journey from rigid system blocks in 2007 to process-driven workflows and generic objects in 2019.
5:40Successful Large-Scale No-Code Implementations
Kowalski showcases real-world examples of complex systems built entirely with no-code, including a comprehensive CRM/SCM app and a university recruitment system.
7:52No-Code versus Code in Integrations
A comparison of how no-code and traditional coding handle data flows, highlighting the strengths of JSON/REST in no-code versus legacy databases and AI-assisted coding in traditional development.
13:26Managing Advanced Logic and Performance
The talk examines the limitations of using no-code formulas for complex math and explains why traditional coding languages offer superior debugging and performance.
16:48Market Predictions and the Impact of AI
Kowalski discusses analyst forecasts from Gartner and Forrester alongside the disruptive impact of generative AI on the developer job market.
20:29Best Practices for Hybrid Application Development
The speaker provides a blueprint for what to build with no-code versus what to program using universal languages and corporate APIs.
24:28Addressing Version Control and Environments
During the Q&A session, Kowalski addresses the difficulties of version control, testing, and environment separation in large-scale no-code projects.

Read edited transcript highlights

These concise notes were edited from automatic captions and checked against the talk structure. They are not a verbatim transcript.

Transitioning to Generic Objects

In their development journey, Archman transitioned from rigid, pre-defined system fields to generic objects. In modern no-code platforms, creators do not start with a predefined set of specific business entities like counterparties or products. Instead, they work with general objects, defining their attributes, forms, and lists from scratch. This flexibility allows users to build highly tailored applications without being constrained by a fixed database schema.

Watch from 4:20

The Reality of Enterprise Integrations

When dealing with large-scale enterprise applications, developers often face legacy systems that rely on XML, database views, SQL procedures, or flat files rather than modern REST APIs. While no-code platforms struggle to handle these diverse and messy data sources, traditional coding allows developers to write custom connectors. Furthermore, generative AI has made writing these complex code-based connectors much easier by generating the necessary code blocks when given clear inputs and outputs.

Watch from 11:16

AI as the Great Equalizer in Programming

Generative AI has fundamentally changed how we view traditional coding. Previously, writing software required deep technical expertise, but today, even non-programmers can build applications by leveraging AI assistants. If you understand the basic logic of what a piece of code is supposed to do, you can ask an AI to write it, test it, and then manually tweak specific lines to fit your needs. This shift has completely disrupted previous market predictions about a permanent shortage of software developers, as AI has dramatically boosted developer productivity.

Watch from 18:41

A Balanced Blueprint for Large Applications

To successfully build large applications, organizations must use no-code with moderation. No-code is ideal for elements that change frequently or require rapid deployment, such as dashboards, user permissions, notifications, and business workflows. However, complex backend calculations and heavy integrations should be coded from the start. When coding is necessary, developers should use universal languages like C#, Java, or Python rather than proprietary platform wrappers, and they should build simplified internal corporate APIs to streamline communication between systems.

Watch from 20:29

The Version Control Dilemma in No-Code

One of the biggest hurdles in large-scale no-code development is version control and merging changes. In traditional software engineering, developers can easily compare and merge lines of text using standard tools. In contrast, no-code platforms rely on visual tiles and graphical workflows, making it extremely difficult to merge a developer's test version with the live production environment. While some platforms offer environment separation, upgrading versions in massive no-code projects remains a significant challenge that often leads to deployment difficulties.

Watch from 26:04

Explore WaysConf 2026

View the agenda

Get tickets