Try to keep up
As I finish the implementation of yet another feature with the help of Claude Code and without having to write a single line of code or tests, I am reminded once more of how fast our profession is changing. My codebase is a few thousand lines of code, mostly C# and Typescript. The feature was pagination applied to both backend endpoints and also frontend tables, so when you navigate to new pages, it requests them from the backend on the fly. I completed the task in under 2 hours and with all of the code, both backend and frontend, covered by tests. I have instructed Claude Code to use my coding style and that’s strict TDD 99% of the time. The side effect is a very well tested codebase.
AI coders love types and they also love feedback given by tests and failing builds(due to, you guessed it, strongly typed languages).
What you know today, may become out of date by the next month. Looking at the skill level of engineers out there there’s a very wide spectrum of AI literacy and AI-coding assistance expertise.
Starting from the developer who may refuse to use any AI tools at all, to the one who crawls Hacker News for the latest developments to this particular area. Do you remember when someone had Visual Studio in their CV as a skill worth making the interviewer aware of? I do. It was a big deal back then. It meant you didn’t rely on Notepad to write your code, it meant that you were efficient at your craft.
Got the analogy yet? Maybe let me try to drive it home.
A few years ago, Github Copilot came out. It was an interesting tool and it slowly captured a good portion of the market. To this day, there are developers who rely solely on Copilot. You can say that an evolution of this is ChatGPT/Gemini/Grok prompting. Perhaps, a successor to Stackoverflow. Here’s my problem, BLAH, give me the solution. Amazing. Once you get hooked to that, if you are curious enough, you might be wondering, OK, but how can I apply this to a codebase? How can I skip the back and forth to my browser and the inherent curse of being a clip board jockey?
Then, you may enter the agentic world and use Claude Code or a similar tool. But even then, there’s depth. Depth that is increasing by the day. From MCP tooling that you can hook up to it, to agentic personas you can create and leverage to do separate tasks … and so on.
The point is, if you are not keeping up with the latest developments in this area, you may find yourself left behind.
Getting Started
If you’re looking to dive deeper into agentic coding tools, I recommend starting with Claude Code: A Highly Agentic Coding Assistant from DeepLearning.AI. This course provides hands-on experience with the tool I mentioned and covers the fundamentals of working with AI coding assistants.
For advanced usage tips and techniques, check out this comprehensive guide that covers more sophisticated workflows and best practices.
So do try to keep up. It’s a brave new world, and it’s moving fast.