Zoho founder Sridhar Vembu has urged software engineers to make use of artificial intelligence while ensuring that they do not lose their understanding of the technology and systems they are building. His comments have sparked a wider discussion about the growing use of AI coding tools and the balance between faster software development and human technical expertise.
Vembu shared his views on X on September 21, 2026, in response to a post from a software engineer who described extensive use of AI across a company's software development process. According to the post discussed by Vembu, artificial intelligence was being used for specifications, code, testing, product requirement documents, tickets, ticket resolution and reports. The engineer also claimed that developers were being pushed to release products as quickly as possible. These claims were made by the individual on social media and were not independently verified.
Responding to the post, Vembu said that he tells engineers at Zoho to use AI but not give up their understanding to it. He also compared the industry's rapid adoption of AI tools to driving a sophisticated new car without sufficiently understanding how to operate it. He warned that moving too quickly without adequate understanding could create problems for software teams.
The comments come at a time when AI-powered coding assistants are becoming increasingly common in software development. These tools can assist with writing and reviewing code, generating documentation, creating tests, analysing errors and completing other development tasks. For engineers and technology companies, such tools can reduce the amount of time required for repetitive activities and allow developers to focus on other aspects of a project.
However, the increasing role of AI has also raised questions about how much technical responsibility should remain with human engineers. If developers accept AI-generated code without understanding how it works, they may have difficulty identifying errors, security vulnerabilities, performance problems or compatibility issues. Human review and technical judgment therefore remain important when AI is used as part of a software development workflow.
Vembu's latest comments are consistent with views he has expressed previously about the changing role of engineers in an AI-driven software industry. In a Zoho interview, he said AI could handle a large amount of routine programming work, but argued that software engineering would increasingly involve using, creating and managing AI tools rather than simply writing code. He also stressed that engineers need sufficient fundamentals to assess the quality of AI-generated output.
In that earlier discussion, Vembu compared AI tools with resources such as Stack Overflow or Wikipedia. Such tools can help people overcome problems or find information, but they are most useful when the person already understands the underlying problem. Without that context, he argued, users may not know whether an answer is appropriate or how to evaluate it.
This distinction between using AI and depending completely on AI is becoming an important issue in software engineering. AI systems can generate code quickly, but generated code still needs to be reviewed, tested and integrated into larger systems. Software development also involves architecture, security, scalability, maintenance, business requirements and decisions that may not be fully captured by a coding prompt.
Vembu's latest remarks specifically focus on the risk of losing technical understanding. His concern is not that engineers should avoid AI. Instead, he has encouraged engineers to use AI as a productivity tool while continuing to understand the systems for which they are responsible.
The discussion also highlights a wider change in the software development industry. AI assistants are increasingly being used at multiple stages of development rather than only for simple coding suggestions. Depending on the organisation and workflow, developers may use AI for generating code, preparing documentation, writing tests, debugging, analysing requirements and handling routine development tasks.
This shift could change the skills expected from software engineers. Basic coding ability may remain important, but engineers may increasingly need strong knowledge of software architecture, system design, security, debugging, testing and domain-specific requirements. The ability to evaluate AI-generated output could also become an important part of technical work.
There are also concerns about less experienced developers. Junior engineers traditionally build expertise by writing code, debugging problems and understanding how systems behave. If AI performs too much of this work from the beginning, there is a possibility that some developers may have fewer opportunities to develop those foundational skills. This issue has become part of the broader discussion about AI's impact on education, training and early-career technology jobs.
At the same time, AI tools can provide significant benefits when used appropriately. Developers can use them to explore possible solutions, automate repetitive tasks, generate initial code and documentation, and accelerate routine parts of the development process. Experienced engineers can then review and adapt the output based on their understanding of the system.
Vembu's message therefore focuses on maintaining a balance between automation and human expertise. His comments suggest that productivity gains from AI should not come at the cost of engineers losing the ability to understand, evaluate and maintain the software they create.
The debate is also relevant to companies deciding how quickly to integrate AI into their development workflows. Organisations may need to consider not only how much faster AI can produce software, but also how AI-generated work is reviewed, tested and maintained. Clear responsibility for technical decisions remains important even when AI is involved in the development process.
The discussion following Vembu's post has included different views from technology professionals. Some argued that AI can help companies and startups move faster, while others focused on the risks of releasing software that developers do not fully understand. The differing responses reflect the wider uncertainty surrounding the appropriate role of AI in software engineering.
For engineers, the debate is ultimately about how AI should fit into their existing technical skills. AI can automate parts of software development, but engineers remain responsible for understanding requirements, evaluating solutions, identifying problems and ensuring that systems work as intended.
Vembu's latest warning does not call for engineers to reject AI. Instead, he has encouraged them to use the technology while retaining technical knowledge and judgment. As AI-assisted software development continues to expand, the ability to combine AI tools with strong engineering fundamentals is likely to remain an important part of the ongoing industry discussion.