George Bernard Shaw once said that progress is impossible without change, and those who cannot change their minds cannot change anything. It is hard not to agree with these wise words. In recent years, AI technologies have become a driving force in the programming world, including web development. It is fascinating how AI-powered tools have evolved from a super-hyped phenomenon to a real game-changer in both the development process and functionality of web applications.
AI is becoming one of the most promising growth directions for IT companies. So, you may be wondering how AI is used within DHTMLX? We surveyed the employees on the state of AI inside the company and found some interesting insights. Our questionnaire covers a range of AI-related questions, from general considerations to developer-specific topics.
In this blog post, we’ll analyze the survey results and see how they correlate with popular surveys focused on AI.
How the DHTMLX Team Uses AI Tools in Daily Work
The concept of artificial intelligence was known for many years without creating too much rush, but everything changed with the arrival of generative AI tools such as ChatGPT, GitHub Copilot, Gemini, Claude, etc. These tools bring true and tangible value to IT specialists in many companies, including DHTMLX.
Most participants of the DHTMLX AI survey are developers, marketers, and tech support specialists. With the current level of advancement, AI tools prove to be very helpful in these particular areas of work. Therefore, it is not surprising that DHTMLX specialists see the benefits of using AI tools on a regular basis (daily or weekly).
To be more specific, the DHTMLX team uses AI tools to simplify/facilitate various tasks, from working with text and images to generating and reviewing code. But the main goal of using AI at our company is research and development.
Although AI instruments have already become helpful assistants of the DHTMLX staff, the overall AI satisfaction rate in web development among the respondents is 3.57 out of 5. Also, there is no consensus on whether AI tools make the DHTMLX team more productive or not. What are the reasons for divided opinions on artificial intelligence? Let’s find out.
Challenges Faced by DHTMLX Specialists When Using AI
Many of today’s most popular AI tools (ChatGPT, Google Gemini, Perplexity, etc.) are relatively new. Even though they already have impressive functional capabilities, AI tools are not flawless.
Respondents of our survey pointed out multiple shortcomings they identified in AI tools and AI-generated content.
Apart from current drawbacks and inconsistencies, members of our team also shared their thoughts on features that are missing now in AI tools:
Note: Some of these features may already be available in lesser-known AI tools, but the suggestions of our specialists refer to more popular AI tools, with which our colleagues have a greater user experience.
- Reusable templates
AI tools are often used for performing routine and repetitive tasks, so it would be very convenient and time-saving to be able to store popular prompt patterns as templates. Instead of retyping specific instructions for artificial intelligence each time from scratch, templates will do the job for you.
- Conversation branching
This feature can become a game-changer, especially for complex topics, providing the ability to fork a discussion with AI into multiple directions without losing previous context. Thus, it gives a great opportunity to consider multiple solutions for a specific question and choose the most appropriate one.
- Stronger contextual and abstract reasoning
For now, AI chatbots can comprehensively respond to detailed prompts. But once you omit a piece of relevant data, AI tools often struggle to fill in the blanks, which eventually leads to errors. Therefore, another possible direction to strengthen AI tools is to enhance their understanding of general and abstract requests.
- Deferred actions
Generally speaking, modern AI tools serve as Q&A systems i.e. they generate content only in response to queries (prompts). The ability to perform deferred tasks such as reminders (“review the report before submitting it”) or recurring actions (“summarize project updates every Friday”) will enable AI tools to operate as proactive personal assistants.
- Self-review and authoritative sources of information
Since AI tools are frequently prone to errors and inaccuracies of different scales, it would be nice to have some control mechanisms to reduce the likelihood of receiving invalid data. For instance, there may be an optional review mode that automatically cross-checks data generated by artificial intelligence. Another possible measure to enhance the reliability of AI tools, specifically chatbots, is to make sure their responses are based on up-to-date and trustworthy sources of info by default.
That’s it with the general considerations about AI at DHTMLX. Now, it is high time to proceed to the most interesting part of the story, namely, the impact of artificial intelligence on the coding experience of our developers and tech support specialists.
AI Usage in DHTMLX Development Workflow
Like many other software development companies, we see that the reasonable use of AI technologies can streamline the development processes. Many members of our development and support teams, who participated in this survey, use popular AI coding assistants such as GitHub Copilot (59,1%), Claude Code (13,6%), JetBrains AI (13,6%), Cursor (9,1%), Codeium (9,1%), etc., to achieve coding goals of different complexity levels.
But as mentioned earlier, AI tools currently have some imperfections, which are applicable to AI coding assistants as well.
In the code written by most of the DHTMLX programmers, the amount of AI-generated code does not exceed 25%.
The main reason for this lies in the fact that our devs have to rewrite from 50% to 100% of the code received from AI assistants. They mentioned multiple reasons why they have to rework the AI code before adding it to a real project.
We also touched upon a new, controversial approach to software development called vibe coding, which implies the use of prompts in LLMs for delivering a complete app’s codebase. We haven’t seen much enthusiasm about vibe coding in our team. Only a few people at DHTMLX played around with this development practice, but it is not considered an option for serious tasks.
There are some real concerns at DHTMLX that overreliance on AI tools may have a negative impact on the professional skills of IT specialists, or even pose a threat to humanity.
These findings lead us to the main conclusion that AI tools are great auxiliary instruments, but in no case can they fully substitute the activity of real specialists. The participants of our survey speak in favor of further exploring and using artificial intelligence in practice.
However, it is also necessary to adopt a sensible approach to AI technologies, considering their current capabilities and limitations. AI tools are suitable for brainstorming and routine work, but even common tasks require time for human verification, let alone more complex scenarios. With AI tools, the devil is also in the details, meaning that the issues in tasks handled by AI are often hard to detect. Modern AI tools are impressive, but they are not a magic pill for care-free coding.
Conclusion: Balancing AI and Human Expertise
All in all, the views of the DHTMLX team on the use of AI tools in the IT industry, and particularly in software development cycles, largely correlate with other surveys covering the topic, such as State of Web Dev AI 2025 and Stack Overflow Developer Survey 2025. Despite the growing adoption of AI tools, IT professionals are not ready to delegate to AI serious tasks that determine the success of real projects.
We cannot accurately predict how artificial intelligence may further affect web development in the coming years. But it is here to stay for a long time, and we should make the most of AI’s current benefits right now to remain competitive in the future. We are already cautiously experimenting with the use of artificial intelligence capabilities in our JavaScript Suite widgets. The first results are two demo apps – AI Form Builder and AI-Powered Customer Review Analysis. And there is more to come. Stay tuned.