In a revelation that signals a potential paradigm shift in software development, Yonatan Levin, a former top executive at the popular project management platform Monday.com, has announced that his new startup operates with artificial intelligence writing 100% of its codebase. After spending six years at Monday.com helping build one of Israel’s most successful tech companies, Levin has ventured into uncharted territory with what he describes as a ‘secret mode’ operation that could redefine how software companies are built from the ground up.

The announcement has sent ripples through the tech community, raising fundamental questions about the future role of human programmers and the viability of AI-first development approaches. Levin is scheduled to share more details about his groundbreaking venture at the Lviv IT Arena conference in September, where industry professionals will have the opportunity to learn firsthand about the practical implications of running a company where artificial intelligence handles all programming tasks.

The Rise of AI-Powered Software Development

The concept of AI writing code is not entirely new, but Levin’s claim of achieving 100% AI-generated code represents a significant leap from current industry practices. Over the past few years, tools like GitHub Copilot, ChatGPT, and Claude have increasingly assisted developers in writing code snippets, debugging, and automating repetitive tasks. According to recent industry surveys, approximately 70% of developers now use AI coding assistants in some capacity, but these tools typically handle only 30-40% of actual code generation, with human oversight remaining essential for complex logic, architecture decisions, and quality assurance.

The evolution of AI coding capabilities has been remarkably rapid. In 2021, GitHub launched Copilot, which quickly became the most widely adopted AI programming assistant. By 2023, major tech companies including Google, Amazon, and Microsoft had integrated AI coding tools into their development workflows. Studies from Stanford University and MIT have shown that AI-assisted programming can boost developer productivity by 55% for certain tasks, while reducing the time spent on boilerplate code by up to 80%. However, critics have consistently argued that AI cannot fully replace human intuition, creativity, and the deep understanding required for building robust, scalable software systems.

Implications for the Tech Industry and Workforce

Levin’s approach raises profound questions about the future of software engineering as a profession. The global software development market employs approximately 27 million professional programmers, with the number expected to reach 45 million by 2030. If AI can truly handle all coding responsibilities, the industry could face unprecedented disruption. However, many experts suggest that rather than eliminating jobs, AI will transform them. Developers may shift from writing code to becoming AI supervisors, system architects, and quality controllers who guide and validate AI-generated solutions.

The economic implications are equally significant. Traditional software development is expensive, with average developer salaries in the United States exceeding $120,000 annually. Startups typically spend 60-70% of their initial funding on engineering talent. An AI-first approach could dramatically reduce these costs, potentially democratizing software development and enabling smaller teams to build products that previously required dozens of engineers. This could accelerate innovation while simultaneously creating new categories of jobs focused on AI training, prompt engineering, and human-AI collaboration.

Challenges and Skepticism Surrounding Full AI Automation

Despite the excitement surrounding Levin’s announcement, significant skepticism remains within the developer community. Security concerns top the list of potential issues, as AI-generated code may contain vulnerabilities that human reviewers might catch. Research from security firm Snyk found that AI coding assistants can introduce security flaws in approximately 40% of generated code if not properly supervised. Additionally, questions arise about intellectual property, code originality, and the ability of AI to understand nuanced business requirements that often require extensive human communication and interpretation.

The ‘secret mode’ nature of Levin’s operation has also raised eyebrows, with some industry observers questioning whether the approach can scale beyond early-stage development. Complex enterprise software often requires integration with legacy systems, compliance with industry regulations, and customization that demands deep domain expertise. Nevertheless, Levin’s willingness to publicly discuss his methods at a major technology conference suggests confidence in the model’s viability. His presentation at Lviv IT Arena is expected to draw significant attention from investors, entrepreneurs, and developers eager to understand whether this represents the future of software creation or an ambitious experiment with limited applicability.