If someone had told me that a computer could write code for me, I would have laughed just two years ago. But here I am, truly astonished by the sheer amount of work we can get done by adopting Artificial Intelligence in software development. That’s what prompted me to write this piece on AI in Software Development.
From Initial Skepticism to Daily Reliance
If you are close to Salesforce development, you know that even though you can code a lot of features, not everything is standard – the framework is like a proprietary version of Java (Apex) combined with HTML, CSS, and JavaScript (LWC/Aura). So, I naturally assumed it would be a bit difficult for Generative AI tools like ChatGPT, Gemini, and Claude to master. At least, that was my understanding from using the early versions of ChatGPT that I had access to during my time at Gartner. Most of the time, it simply could not provide me with proper solutions to the specific problems I was trying to resolve. It hallucinated frequently and made numerous mistakes. Honestly, it was a pain trying to explain the issues and getting the AI to accurately understand the problem statement. Meanwhile, others were raving about how easy it had made everyone’s lives. I just didn’t realize what I was missing back then.
That is exactly why I was initially hesitant to use Copilot at my current organization. However, working at a company that likes to stay ahead of everyone in the game, it felt like I was quickly becoming outdated without using it. So, I started adopting it slowly. First, I used it for finding answers to general issues related to everyday Salesforce problems that I would typically use Google Search for. But now? Now, it is beginning to look like I absolutely cannot live without ChatGPT or Claude at work – whichever model GitHub Copilot happens to choose based on availability inside VS Code.
Every single new model is growing by leaps and bounds and becoming extremely intelligent. Salesforce, which once struggled to get accurate answers from these tools, has been thoroughly mastered by AI. Not a single day goes by without me using at least one of these Gen AI tools. It is able to easily understand Apex, LWC, and most of the other complex Salesforce metadata and relationships.
Doing More in Less Time
On one side, I am incredibly happy that it has made life easier for software developers like me. I am able to do more in less time and can deliver an even higher volume of work in the usual time that I have. Because of this newfound efficiency, I am able to start learning and working on other Salesforce platforms, like Data Cloud (Data 360) and Salesforce Marketing Cloud, where the effect of AI isn’t quite as impactful as it is on the core Salesforce Platform yet. For people who aren’t familiar with the Salesforce ecosystem, there are a lot of different versions – or “clouds,” as they like to call them offering different sets of features specific to the industry they are targeted toward. For example, there is Sales Cloud for sales representatives and Life Sciences Cloud for the healthcare industry.
The Future Workforce: Developers vs. Prompt Engineers
On the other hand, I am genuinely worried that fewer people will be required overall to perform the exact same amount of work. We have already come across news where non-tech-related roles are getting rapidly eliminated and replaced by AI. AI chatbots have already taken over customer support across almost all platforms, and finding a real human to get help from is quickly becoming a rare sight. Now, the heat is turning toward skill-based jobs like software development. Although Gen AI isn’t truly at its absolute peak yet (we all know that AI is rapidly evolving and each new model is significantly more capable than the one it replaces), it is now at a place where five people can easily do the work that previously required the same amount of effort spent by ten to fifteen people.
In the near future, we might not need more traditional coders, but rather prompt engineers – someone who is able to accurately describe the complex requirements to the AI model so that it can produce a working application in just a few seconds. This shift is either going to completely eliminate traditional developers or business analysts for sure. Let’s see who is quick enough to adapt to AI and make it work to their advantage.
I’m betting on the developers!
Discover more from Antony Agnel
Subscribe to get the latest posts sent to your email.