Init: Navigating generative AI in software development

Since entering the tech world, one question has consistently intrigued me: how can we enhance tech workers’ efficiency and accelerate adoption? The pace of digitalization has always seemed to scale linearly, falling short of the exponential growth many tech enthusiasts envision. In my assessment, the limited number of software engineers we could produce and train has been a significant bottleneck, along with more or less linear productivity scaling.

During my studies and graduate research, I encountered generative programming as a potential solution for scaling challenges. I still fondly remember that excitement. For a period, I distanced myself from cutting-edge tech research and focused on products and consumer-focused software. Then, in November 2022, OpenAI released GPT-3, which among many other things, felt like a milestone in software engineering tools. This event resonated with me in a way similar to other transformative technological moments I’ve experienced: my first internet connection in late 90s, my first smartphone (a Samsung S1), and now, my first interaction with GenAI. That pivotal moment occurred while exploring lex.page, a writing tool I discovered through my every.to subscription. The experience of typing ‘+++’ and seeing the next sentence(s) appear was truly remarkable. It felt like a collaborative writing process with an AI that could anticipate my thoughts. This interaction sparked a renewed interest in AI for me, and before long, reignited my interest in software development tools with a new twist of GenAI impact on those.

After a year and a half of daily AI-assisted development and actively thinking about how to improve my workflows, and also the team productivity, I felt it would be a good idea to start sharing, reflecting on it, and speculating a bit about the future.

Who am I and why is this relevant?

Now, you might be wondering, “Who is this guy, and why should I care about his GenAI opinions?” Fair question. Let me give you a quick run-down of my journey:

My programming journey began at 12. This led me to study in Kyiv, and later to a scholarship to study at Purdue University. After graduation, I joined Microsoft, working on developer tools for Microsoft Office engineers. This gave me valuable insights into how software development can be organized at that enormous scale. Then, my entrepreneurial itch led me to various consulting projects and startup attempts, including a mobile app for GlobalID and an emissions estimator for ClimateChange.org. Over the years, I’ve delivered or supervised the development of more than 20 different products, primarily focusing on early-stage startups. Currently, I’m the CTO at WERZ, a company I joined full-time in 2022 after initially building their development team as a consultant.

The relevance of this background to GenAI, particularly developer tools, is significant. My varied journey, encompassing experiences in big tech, consulting, and startups, has provided me with a unique blend of perspectives on technological evolution. I’ve gained insights into how new technologies can impact software development, especially in iterating on early ideas and bringing products quickly to market. I’ve observed various hype cycles in the tech industry, such as the ML craze of the late 2010s when we thought fully self-driving cars were just around the corner, and the crypto/blockchain cycles, notably during the ICO mania of 2017 and the subsequent cycle in 2021–2022. The current Generative AI hype feels different, yet also shares some similarities.

I strongly believe that GenAI is transformative. However, in its current state, it is over-marketed, with extreme takes prevalent — ranging from “GenAI is gonna be useless” to “GenAI will replace X in Y years”. My perspective is akin to how people describe the internet pre-dot-com bubble burst — the technology feels transformative, yet it will take time to fully feel the impact. To me, it’s a powerful tool with significant capabilities, but it’s not a magical solution to all development challenges. However, on balance, it has already brought tangible results. I don’t think we could have delivered as much as we did over the past 1.5 years (2 mobile apps, 2 web apps, and infrastructure to tie it all together) without it.

Join the AI-Assisted developer journey

In the upcoming articles, I’ll share more of my hands-on experiences, both the wins and the stumbling blocks. We’ll look at practical applications, like how AI helped streamline the process of building a Figma-like editor from scratch. But we’ll also dig into the challenges, especially when dealing with more complex tasks or niche technologies.

My goal is to give you a real-world perspective on GenAI in software development. I will show you where it shines, where it falls short, and how to navigate its current capabilities and limitations. Whether you’re deep in the coding trenches or just curious about where tech is headed, I hope these insights will be useful.

I’ll be posting new articles regularly, aiming to cover a range of topics related to software development with focus on GenAI, but not limited to it. If you’re interested in this intersection of AI and software development, stick around. Let’s figure out how to make the most of this technology one code suggestion at a time.