What AI-assisted software development really feels like (spoiler: it’s not replacing you)

Coding isn’t the same as it used to be. AI systems can write code with a single-shot prompt. 

While some fear that it is going to take away the jobs of developers, many argue that it is more like a “pair programmer” than anything else. 

On a post by Business Insider in December 2024, 

Amazon Web Services said that developers report spending an average of "just one hour per day" on actual coding. The rest of the time is spent learning codebases, drafting documents, testing, overseeing releases, fixing problems, or hunting down vulnerabilities. 

While the “one hour” might be a bit of an exaggeration, engineers indeed spend a big chunk of time doing a ton of other tasks besides just coding, and AI performs fairly well at assisting you in coding the repetitive stuff, especially since it is trained on public data. 

So, your chances of being replaced by AI are out of the window. 

Our focus should be on upskilling in AI rather than seeing it as a threat. I draw the analogy to the Cloud; back then, people were scared about losing jobs to the Cloud, but now it ended up creating more than before.

 There are endless debates on this, which we can park for now, but instead, we can talk about AI in software development in this article. 

Before we get started, I tweeted this a few days back, and some of the responses I received were along the lines of “I think it's going to be another tool; a skilled craftsman knows how to use it, one who doesn't slowly disappears.” 

I would also like to know your thoughts on this before you scroll down to hear my views on AI-assisted development.

What is AI-assisted software development, and why should you care?

There are multiple definitions of “AI-assisted software development.” 

I see it as a “buddy” who is right there with you in your IDE, assisting in different parts of software development – coding, writing PR messages, documenting, and researching. 

While there are many reasons why you should use AI for software development, one simple reason is that developers around you are doing it, and companies are adopting AI-assisted software development. 

If you don’t learn how to code with AI, you are not upgrading and might eventually fall behind your peers when it comes to shipping, even if it’s a small think like coding with ChatGPT. 

Additionally, there are multiple reports on how AI-assisted coding has helped companies have faster releases and better cater to customers, along with a report from JPMorgan where they mentioned that AI has led to a 20% increase in productivity among developers.

But today, we will talk about something that is important and less talked about, and that is burnout/mental fatigue. 

How AI can help in reducing burnout

If you are active on social media, you have probably heard about “touching grass.” 

If not, it simply means making sure to take breaks between work. 

This is especially true in tech jobs, where we spend more time in front of our computers than outside, leading to mental health issues (a very hush-hush topic in tech). 

According to research by the Society of Research Software Engineering, 51% of software engineers have been diagnosed with a mental health condition, 57% reported burnout, and 71% said their productivity was reduced by a mental health issue. 

Additionally, there are many developers who want to quit their jobs and become farmers (this is not just a meme; there are multiple people in IT who have moved to becoming farmers, if you remember the goose farmer story).

I personally believe that AI can solve this to some extent. 

And here’s where developer retention quietly enters the picture: Burned-out developers don’t just ship less – they leave.

By reducing repetitive tasks and helping devs focus on meaningful work, AI-assisted workflows create space for fulfillment and flow – two things that directly impact retention.

Here’s a comment on Reddit from a developer who mentioned that their mental health worsens when they work on not-so-challenging tasks, and multiple developers agreed with this point. 

This brings me back to my original point of AI helping in reducing burnout. Since it can automate the mundane parts of your job, you can spend more time doing challenging work and work that truly makes you happy.

Here are some ways that I feel AI can bring back the adrenaline rush to our dev jobs:

More focus and less context switching

Most of us want to write more code and spend more time working on complex solutions and thinking about how we can solve problems.

A major blocker in the development process was the context switching in our jobs – having to read documentation, read and understand legacy code, and spend weeks just looking for resources. 

This is a major help that AI has provided. 

We no longer need to spend long hours just to understand code and close 100 tabs at the end of a task.

Here’s an example of me using AI with VSCode to understand some blocks of code. (It still surprises me that I do not need to Slack someone or get on a call when I don’t understand something.

When developers can stay in the flow and avoid constant disruption, not only does productivity improve – but so does satisfaction. This kind of frictionless experience makes them far more likely to stay, grow, and evolve within a team instead of burning out and moving on.

AI has reduced dependency for me and now helps me move faster.)

AI as a buffer against cognitive overload

Similar to context switching, developer burnout also comes from the mental effort required to track everything – requirements, dependencies, system constraints, team decisions, and even things like status reports.

I have personally been through instances where I missed updating a sheet or reporting what I worked on, and I wished I had a tool that could track everything I did and summarize it for me, basically acting as a second brain.

Now that AI has context preservation, I can simply add links, documents, files, and code as context and then ask AI to work on it. These are some of the things we did not actively "count" when working, but now that AI is handling it for me, I realize the amount of work it took previously.

For example, I write a lot of blogs, and sometimes I face writer’s block or lack ideas on what to write about next, so here’s me using my blogs as a context and asking AI to analyze what I write about and suggest topics based on it.

Using AI as a copilot and focusing more on architecting

There have been tweets from people saying, “AI will force developers to level up on architecture. When AI handles the grunt coding, the human job becomes designing systems that don’t collapse.” 

This comment is also reassuring in that our jobs are safe even when AI is capable of writing code. 

I recently started working on infrastructure stuff at work, and it is a lot of fun. Knowing how to architect software that can be scaled to millions is the fun part of being an engineer to me.

Here’s a tweet from a dev who mentioned that Cursor made him fall in love with coding again while he was suffering from burnout. 

I am curious to learn if there’s any such tool that has helped you. For me, using Pieces as a staple for development and other tasks has helped me reduce context switching by a lot. 

Oh, and by the way, you could also integrate it with Cursor.

AI as a retention strategy (yes, really)

Most conversations about AI in software dev are about speed and cost – but what if it could help you keep your best engineers longer? 

❗Repetitive tasks, cognitive overload, and lack of creative work are all major contributors to developer churn.

By offloading grunt work and preserving context, AI helps devs spend more time doing what they love: solving interesting problems. 

So why not avoid context switching now? 

For example, I am saving snippets of code and documentation which I frequently refer to, this saves my time in going through multiple bookmarks and finding that one solution that worked for me.

I even took a step further and used long term memory to capture ai, kind of to shadow me while I work, remember the files I worked on, the sites I visited, and the code I wrote, so that it can be used as context for the other tasks I will work on (if you are worrying about security, you can use local models, so everything that you do is on your machine).

Since I had the LTM (long-term memory) turned on, Pieces, using its workflow history feature, summarized everything that I worked on and the next steps that should follow. 

This not only helps me get up to speed but is a great help when you start working on something after a long time (or maybe you went on a vacation and no longer remember what you worked on, this is classic me, so it is a BIG PLUS for me)

Lead smarter with AI-assisted software development

Do you know how most of these articles talk about how AI can help developers but never focus on how leaders can use it? 

Well, let's change that. 

Even though I am not in a “leadership” position, I observe a lot! 

I will be writing some points based on my observations. 

First things first, as a leader, you are not only responsible for managing your team and your KPIs but also for ensuring that everyone is happy doing their job. 

Empowering your team to do more high-impact work is a great way to make them stand out in a competitive market like this, and you, as a leader should enable that.

But how does it correlate with AI-assisted development? 

Taking up new and more challenging tasks 

We often end up doing tasks that we are comfortable in; for example, since I code in JavaScript, there were fewer chances that I would pick up a task in Rust (especially at work where we are bound by timelines), but AI has not made it faster by reducing time to production. 

I can get my hands dirty with code in just a few days after learning the basics since AI can help me research, write, and understand code. 

Similarly, you can diversify your team’s skill sets by enabling them to work faster with AI.

Building applications that are “security first”

While all the applications we build should be “security first”, many products are guilty of treating security as an afterthought. 

With the increase in artificial intelligence programming, security is an even bigger element in software development than before. 

Recently, we have seen vulnerabilities from Next.js and the famous “vibe coded SaaS tweet” (screenshot below) that didn’t age well.

 While AI is capable of writing a lot of code, it is not able to write code for building a scalable and secure application, and this is where more focus will be in the upcoming days. 

You have the perfect chance to enable your developers to automate the mundane parts of the job with AI, or open even some positions that will only gain traction while letting them focus on more critical aspects like security. 

This is a win-win for you as a leader and also for the company. with more focus on security, you are likely to gain trust from devs.

Ending thoughts

The tech industry is moving fast, and we all know it, but what we don’t know is what the future looks like (I mean what the next 6 months to a year look like, considering how there is a new model and AI capability every week). 

With the discussions that I participate in, and from tweets from famous tech founders, I have a feeling that design, product sense, and security are going to be the top 3 areas of focus and not just on “coding”. 

So my advice to you is to embrace the change, use AI tools to automate and “work smarter,” and see how we can make it a part of our workflows.