Generative AI is rapidly gaining traction across industries, and software engineering is often cited as one of the fields poised for the most disruption. Personally, I’ve been slow to adopt GenAI tools in my day-to-day work. This post is a reflection on how I’m currently using them and where they’ve made an impact.
ChatGPT & Perplexity: My New Google and Stack Overflow
Until recently, Google Search and Stack Overflow were my primary tools for finding answers and understanding technical concepts. If I ran into a problem, I would Google it, open multiple tabs, and sift through articles or forum posts until I found what I needed. These days, that’s rare. I hardly visit Stack Overflow anymore, and my Google searches have dropped significantly. ChatGPT has replaced both for most queries.
With Google, finding a clear, complete answer often meant visiting four or five pages. Although Google has introduced AI-powered snippets, I’ve found them factually unreliable in many cases. In contrast, ChatGPT provides a concise, accurate answer in a single response, eliminating the need for link-hopping. That alone has been a major productivity boost.
I also have access to Perplexity Pro through my Airtel subscription. I mainly use it on my phone for quick lookups. Compared to ChatGPT, Perplexity often surfaces different perspectives or sources for the same query. Sometimes its results are better; other times, ChatGPT wins. I haven’t picked a clear favorite yet—they complement each other well.
Copilot & Cursor: AI for Coding
I tried GitHub Copilot at both my previous and current companies but didn’t find it particularly valuable beyond basic autocomplete suggestions. Eventually, I stopped using it.
Recently, I started using Cursor at my current job, and it has been a different story. Cursor does a great job explaining unfamiliar code blocks and even drafting changes. It’s not perfect. Sometimes it stalls for minutes or fails to handle complex requests. But for small tasks, it’s fast and effective. I also use it to brainstorm design ideas or discuss code improvements.
OpenAI API: Programmatic Access to GPT
I’ve set up a prepaid subscription to the OpenAI API, mainly for learning and experimenting with large language models. I don’t use it heavily, but it comes in handy for coursework, prototyping, and automation experiments. Depending on the project, I’ll either call the API directly or use it through the litellm library for easier integration.
Conclusion
The adoption of GenAI tools has definitely changed how I work, turning once labor-intensive searches and coding tasks into much more streamlined experiences. While I’m still exploring which tools fit my workflow best, it’s clear that GenAI is already making a substantial impact on my productivity and learning.
I drafted the post with data points and asked ChatGPT/Cursor to refine it and both did an amazing job. But when I asked for creating an image for the post, Perplexity did a much better job than ChatGPT. It may be due to the fact that I am using the pro version of Perplexity while free version of ChatGPT.