“AI Might Take Away My Design Job” — How That Anxiety Led Us to Rethink the Role of Web Designers

“Now that AI can create banners and landing pages, will designers still have work?”

Since we started using generative AI more often in production, we have been hearing questions like this more frequently.

Working web designers have shared concerns such as:

“If AI can generate ideas faster than I can, where does my value remain?”

People who are just starting to study web design have also asked:

“Is it still worth learning web design now? Will it lead to work in the future?”

These concerns are not exaggerated. In reality, AI can now produce initial drafts of banners, landing page layouts, copy, and image ideas in a very short time. Work that once took hours can sometimes take only minutes.

At ALAKI, our web production and operations team has also been using AI extensively in client projects. While handling web production and operations ourselves, we also develop and provide MONJI+, a web operations support platform built from insights gained in the field.

Through that experience, one thing became clear to us:

The role of web designers is not disappearing. It is changing.

In our own projects, less than 20% of AI-generated drafts were ready to deliver as-is. More than 80% required designers to revise, combine, and refine them before delivery.

So what is expected of web designers in the age of AI?

In this article, we will look back at the mistakes we made while using AI in real projects and explain how we now define the new role of web designers: shifting from execution to final decision-making.

---

Why This Problem Happens

While AI makes design production faster, it also creates new challenges in the field.

The key issue is that there is still a major gap between what AI can generate and what a team can actually use.

AI can quickly generate visually polished drafts of banners and landing pages. However, it does not automatically judge whether the output fits the client’s brand, follows industry norms, or feels trustworthy to users.

Background 1: AI Can Produce Good-Looking Drafts, But It Cannot Fully Understand Brand Context

One of our first major rounds of revisions happened in a project where we submitted a design that used AI-generated copy and images almost as they were.

The client’s response was:

“This doesn’t feel like our brand.”

AI can create polished outputs based on past design patterns. It can make the colors, layout, and tone of the copy look reasonably well organized.

However, it did not capture the subtle nuances of the client’s preferred wording or the way they used color.

In another project, we used AI to build out a landing page almost to the point of release, only to realize just before launch that it did not align with industry norms. We had to rework it.

Because AI learns across many industries, it does not always reflect the expressions, taboos, or unwritten rules that matter within a specific industry.

In other words, AI is good at creating something that looks right.

But judging whether something feels authentic to a company, earns trust in a specific industry, or feels natural to the intended user still requires human understanding of context.

Background 2: Production Alone Makes It Harder to Explain Results

Another challenge is measuring results after launch.

In the past, even when we delivered a visually strong design, there were many times when a client would ask:

“So, did this actually work?”

And we could not answer clearly.

Because we had not prepared a system to measure results after publication, it was also difficult to connect the work to the next improvement proposal.

As AI speeds up production, clients no longer evaluate work only by how quickly it was made.

Instead, questions like these become more important:

・Does this message fit the brand?
・Is this landing page contributing to results?
・What was the reasoning behind this revision?
・Where should we start improving next?

To answer these questions, we need to look not only at the deliverable itself, but also at the decision-making process and post-launch data.

In the age of AI, web designers are not only expected to make things by hand.

They are expected to review AI-generated ideas, discuss them with clients, and take responsibility for final decisions.
That is where the role is changing.

---

Steps Toward a Solution

So where should production teams begin?

Rather than rejecting AI, we began by creating a process that makes AI-generated outputs easier for humans to evaluate.

The key is not to treat AI-generated work as a finished product.

It should be treated as a draft that needs to be reviewed.

STEP 1: Visualize the Current Situation

The first thing we did was visualize which parts of AI-generated drafts were being revised by humans.

In our operations work, we increasingly use a range of generative AI tools to create banners, landing page rough drafts, and first drafts of copy.

When we tracked the results, we found that less than 20% of AI-generated drafts could be delivered as-is. More than 80% required designers to revise and combine them before delivery.

The revisions mainly included:

・Adjusting the tone to match the brand.
・Replacing elements to align with industry norms.
・Correcting inaccurate information.
・Restoring consistency across designs.
・Checking whether the message felt trustworthy to users.

The important point here is not to conclude that AI is useless.

In fact, AI has reduced the time needed to create drafts. However, the time spent on meetings, aligning revision directions, and verification has increased.

In other words, designers’ working hours have not simply decreased.
A more accurate way to describe the change is that the work has shifted from production to dialogue and judgment.

By making this change visible, it became easier for us to redefine the role of designers.

STEP 2: Record and Organize Decisions as Part of Monthly Operations

Next, we began recording decisions about AI-generated work for each project.

If every AI-generated output is simply revised on the spot, the same mistakes are repeated again and again.

For example:

・Words that should not be used for a certain brand.
・Patterns that AI previously generated and the client rejected.
・Industry-specific unwritten rules.
・The tone and nuances that the client values.

If this information remains only inside an individual designer’s head, it cannot be carried over to the next project or shared with another team member.

So we began using the feedback and Wiki features in MONJI+ to keep records of our decisions on AI-generated outputs within each project.

For example, we share an AI-generated landing page rough draft on MONJI+ and leave comments directly on the captured screen.

・“This copy feels off-brand.”
・“This visual doesn’t align with industry norms.”
・“This wording could mislead users.”

By leaving the reasons for revisions directly on the screen, the decision-making process becomes part of the project record.

We also use the Wiki feature to collect project-specific rules and past decisions.

When asking AI to generate the next draft, we can adjust the prompt while referring to past decision logs. New team members can also understand what was rejected and what was approved in the past.

The important point here is not to let designers’ judgment remain an individual, subjective sense.

By recording, organizing, and reusing those decisions, a designer’s value becomes an asset for the entire project.

STEP 3: Use Several Months of Results to Support Discussions and Proposals

The final step is to include post-launch results and connect them to the next proposal.

In the age of AI, web designers need to do more than refine drafts. They also need to look at what happened after the work was released.

In our own workflow, we changed our operations so that, after launch, we could use Google Analytics integration to review performance data in the same place where client revision requests and project communication are managed.

This made it easier for designers and clients to check questions such as:

“Is this initiative actually producing results?”
“Where should we look first for the next improvement?”

Of course, not every initiative produces clear results immediately. Using AI does not guarantee better performance.

Even so, by keeping revision discussions, decision logs, and post-launch data in the same project space, we gradually build the evidence and context needed for the next proposal.

For designers to move from finishing the deliverable to improving outcomes together, this accumulation is essential.

---

What Actually Changed

After we changed our workflow, the role of designers in our team began to shift clearly.

Previously, there were many cases where designers only handled production, delivered the finished work, and ended their involvement there.

But once AI became more common, that approach made it harder to create value. If AI can generate drafts, it becomes difficult to stand out simply by being able to make something look good.

So we started reviewing AI-generated drafts together with clients, using questions such as:

・Does this fit the brand?
・Does this match industry norms?
・Does this message feel trustworthy to users?
・Based on post-launch results, where should we improve next?

Through this process, designers began to act as final decision-makers who identify AI’s mistakes and adapt the output to the brand.

・AI may generate messaging that sounds convincing but is wrong.
・It may create visuals that are slightly off-brand.
・It may write copy that does not fit the industry context.
・It may create structures that feel unnatural or questionable to users.

Designers identify these issues, agree on revision directions with the client, and take responsibility for the overall design.

As a result, the way clients saw designers gradually changed.

Instead of seeing designers simply as people who create things, they increasingly began to see them as partners who take responsibility for outcomes.

This change happened precisely because AI entered the process.

As AI began handling part of production, human designers became responsible for higher-level decisions and improvements after launch.

---

Important Notes and Limitations

That said, there are also important cautions when using AI in production and operations.

First, introducing AI does not automatically make work easier.

In our own team, when we first introduced AI tools, some designers almost lost motivation because they felt, “Maybe I’m no longer needed.”

There was also a period when designers struggled to shift into the role of directing AI-generated work. Instead, they tried to compete with AI on speed. As a result, quality dropped and rework increased.

We thought AI would free up time, but in reality, we sometimes became busier because we had to spend more time checking and revising outputs.

What we learned is that the idea “AI will make everything easier” has limits.

If there is no one who can properly handle AI-generated output, both quality and results may decline.

Another caution is that when tools are fragmented, decisions are hard to accumulate.

・Feedback are in a chat tool.
・Knowledge is in another management tool.
・Performance data is in an analytics tool.
・Design files are somewhere else entirely.

When everything is separated like this, the decisions designers make are not preserved anywhere. They simply disappear.

If decisions do not accumulate, every new project starts from zero. As a result, designers end up returning to production work every time.

A role that is difficult for AI to replace does not come from a special job title alone.

Dialogue with clients, understanding of brand context, industry-specific judgment, and post-launch verification all need to be continuously recorded and reused.
Only then can designers function more effectively as final decision-makers.

---

MONJI+: A WebOps Platform Born from Real Production Work

As discussed above, the role required of web designers in the age of AI is not limited to creating deliverables.

Designers need to review AI-generated ideas, communicate with clients, record the reasoning behind decisions, and continue looking at results after release.

To do that, production, revision requests, knowledge, and performance data should not be scattered across separate tools. They need to accumulate in the same project environment.

MONJI+, the platform we develop and provide, was created from exactly these kinds of challenges in real web production and operations.

With MONJI+, teams can leave comments directly on screenshots of websites and landing pages, organize revision requests and feedback, store project-specific decisions and rules in a Wiki, and review post-launch numbers through Google Analytics integration.

Instead of treating AI-generated drafts as finished products, MONJI+ helps teams review them with clients, record decisions, and improve the work over time.

For production companies and web teams that want to work this way, MONJI+ can be one practical option.

MONJI+ also runs a co-creation initiative with production companies and business-side web teams to explore better ways of managing web operations together.

In AI-era production and operations, how should we record decisions and connect them to results?

We want to work with people who share these challenges and build more practical ways to manage and improve websites.

---

Conclusion

Generative AI has made it possible to create drafts of banners and landing pages much faster than before.

Faced with this change, it is natural for web designers to worry that their work might disappear.

But based on what we are seeing in our own projects, designers’ work has not disappeared. Instead, the role required of designers is changing.

・They need to judge whether AI-generated output can be used as-is.
・They need to adjust it to fit the brand and industry context.
・They need to make decisions through dialogue with clients.
・They need to review post-launch results and connect them to the next improvement.
・They need to record those decisions and carry them into future projects.

We believe that this accumulation of dialogue and judgment is the role of web designers that AI cannot easily replace.

The more AI can generate, the more important human judgment becomes.

In fact, as AI makes it possible to generate more things, the person who decides what to keep, what to change, and what to carry forward becomes even more important.

Web designers are shifting from execution to final decision-making.

We will continue exploring how to use AI while building web production and operations processes where designers’ judgment is properly recorded and connected to results.