In recent time we hear everywhere about AI capabilities, we observe its expanding exponentially. Many AI fans and mainstream media push the software industry to face a question: Can AI replace developers and testers in building and testing software? While this vision might seem like far away dream to some, recent advancements in AI-powered development and testing tools and the rapid evolution of language models suggest this might be closer than we imagine. However, the reality of AI’s current capabilities and limitations deserves a thorough, evidence-based examination. This is why I decided to run a personal experiment and want to share my results with You. I tried to simulate a role of PO or Founder with just an idea of an app that I want AI for me tu build. The goal is to get a working prototype, have descent unit test coverage generated and MVP set of features using AI as the primary developer. Through a structured one week after work execution for AI driven development process. I explored AI’s capability to handle end-to-end software development, from feature implementation to unit testing, using iOS and Swift as our testing ground. I did not analysed the code itself, have very limited programming skills from studies, not used to much afterwards, and had 0 Swift experience. My role was strictly business driven, asking for features, pushing the code to repo and sharing compilation errors or asking for unit test coverage to get better. I did demo style limited happy path scenario testing only. The experiment provided intriguing results: multiple thousands lines of code was generated, partialfunctionality implementation, some of it working as expected, and not bad unit test coverage. These outcomes reveal both the potential and current limitations of AI in software development. During the presentation we’ll analyse the journey of being a pure Product Owner to becoming more technical, examining how AI’s current context limitations and inability to maintain broader architectural vision affect the development process. The session will explore the evolution of human roles in an AI-augmented development environment, discussing how our focus might shift toward more creative and architectural aspects while AI handles implementation details. Special attention will be given to the implications for testing, where we discovered AI’s tendency to adapt tests to code rather than requirements, highlighting the continuing importance of human oversight. We will also include a forecast for the future what can happen and is it really so far from AI having the capability to implement software, what roles were needed and what could a path for us IT experts if the optimistic scenario will appear one the horizon.