跪拜 Guibai
← Back to the summary

When the Frontend Team Got Cut, a 10-Year Java Dev Stopped Worrying and Learned to Work With AI

After the Company Eliminated Frontend Positions, I, a 10-Year Java Developer, Embraced AI Seriously for the First Time

I used to worry: Will AI replace programmers? Now I worry more: Will programmers who can't use AI be replaced first?

Recently, the company eliminated some frontend positions.

After the news came out, what people talked about most wasn't technology, but anxiety:

This is my 10th year doing Java. To say I'm not anxious at all would be a lie. But after thinking it through seriously, I arrived at a somewhat counter-intuitive conclusion:

AI hasn't replaced me. On the contrary, I've become increasingly unable to live without it.

01｜Our generation of programmers has long been accustomed to being "redefined" by the times

When I first entered the industry, it was still the era of JSP + Servlet. One person wrote pages, wrote interfaces, wrote SQL, and deployed to Tomcat. It wasn't that the term "full-stack" didn't exist back then, but there were few people and lots of work, so everyone was expected to know a bit of everything by default.

Later, the separation of frontend and backend arrived. Frontend got its own engineering system, and backend finally didn't have to stitch HTML together inside JSP anymore.

Later still, Spring Boot became popular. Even later, microservices arrived. Nacos, Spring Cloud, registry centers, configuration centers, gateways, circuit breakers, rate limiting, distributed tracing...

During those years, many Java programmers had similar nights:

Writing business code during the day. Studying why Nacos couldn't register at night. Then checking the Spring Cloud version compatibility matrix. Looking up, it was already early morning.

At that time, some people also said: the tech stack is getting more and more complex, old programmers can't keep up. But what happened later? Those who could learn kept moving forward. Those unwilling to change were left behind.

Today's AI is just another paradigm shift in technology.

02｜When the company eliminates frontend positions, is AI really "stealing jobs"?

I think a more accurate statement is:

AI is breaking down the previously clear boundaries between roles.

Previously, the chain for a requirement looked roughly like this:

Product proposes requirement
   ↓
Frontend makes the page
   ↓
Backend writes the interface
   ↓
Testing verifies
   ↓
Go live

Now? An engineer who understands the business and the system can, with the help of AI:

Requirement breakdown
   ↓
Quickly create a page prototype
   ↓
Define interfaces and data structures
   ↓
Generate partial frontend and backend code
   ↓
Verify the core process

This doesn't mean frontend is unimportant. It means that many things that previously required multiple people and multiple stages of collaboration can now complete the first round of validation faster.

Behind the reduction in positions, what's really changing isn't "frontend has no value." It's that companies are starting to value more:

03｜Now when making technical plans, I first "have a fight" with AI

Previously, when writing a technical plan, I would usually do these things myself:

I still do these things now. It's just that I have an extra "tech buddy" who is always online and never complains I ask too many questions.

For example, when a new requirement comes in, I'll first let AI help me do the first round of deduction:

The answers it gives me may not be directly usable. Sometimes they even look very complete but actually don't fit the company's existing architecture at all.

But it has one particularly great value: It can quickly turn the vague ideas in my head into a plan that can be questioned, discussed, and overturned.

Previously, I might need half a day just to outline a plan. Now, I can complete the first version faster, and then spend time on the truly important questions:

For these, AI can give suggestions. But the person who makes the final decision is still me.

04｜AI can write code, but it can't replace "I take responsibility"

Now, letting AI write code is already a daily operation. Interface skeletons, DTOs, VOs, unit tests, exception handling, utility classes, SQL drafts... a lot of repetitive labor, it indeed writes quickly and tirelessly.

But I'm increasingly clear about one thing: Being able to generate code does not equal being able to generate a stable system.

These problems aren't about "who types faster." They're about how many online incidents you've seen, how many pitfalls you've stepped into, how much real business you understand.

AI lets me write faster. But ten years of Java experience lets me know: Which code can be used directly, and which code can only serve as a reference.

05｜AI Full-Stack Engineer is not a more exhausting title

Many people, upon hearing "AI Full-Stack Engineer," immediately think:

Backend has to write frontend. Frontend has to write backend. Plus understand product, deployment, and AI. In the end, one person does three people's work.

This worry is completely understandable. But I prefer to understand it as: An experienced engineer finally has the tools to extend the boundaries of their capabilities outward.

Knowing a bit of frontend allows faster validation of product ideas. Understanding a bit of design allows explaining plans more clearly. Understanding deployment and monitoring makes going live more stable. Being able to use AI allows leaving time for harder problems.

This isn't returning to the era of "doing everything yourself." It's that after we have stronger tools, we regain control over the entire delivery process.

06｜AI reminds me of the me who studied microservices until the early morning

What surprised me most: AI didn't make me lose interest in technology. On the contrary, it reminded me of myself when I first encountered Nacos, Spring Cloud, and microservices.

Back then, there were no ready-made answers. I would flip through documentation for a long time just to understand a service registration problem. I would watch logs all night to locate a call exception. I would be happy for a long time just because a service finally ran through successfully.

Now learning AI is the same.

These questions also have no standard answers. But that's precisely what makes technology most interesting.

Finally

From JSP + Servlet, to frontend-backend separation; from Spring Boot, to microservices; and now to today's AI Full-Stack Engineer.

This isn't the first time we've stood at a crossroads of technological change.

AI will eliminate some repetitive, standardized, low-value work. But it will also amplify those who are willing to learn, understand the business, can make technical judgments, and are willing to take responsibility for the results.

So, for me, who has done Java for ten years:

AI is not here to replace me. It's here to pull me out of repetitive labor and let me refocus my energy back onto solving problems.

And I've finally found again the me who would research a new technology deep into the night.

Do you think AI will change frontend first, backend first, or the entire collaboration model of software development?