Here at Mercari, most members have already started using AI in their daily work as we move toward becoming AI-Native. But how are Mercari’s AI-Native members actually utilizing AI? To uncover the answer to this question, Mercan has launched this interview series titled “Yurino’s AI Question Corner,” featuring Yurino Horiuchi (@Yurino)—a new graduate working on Mercari’s AI Task Force (a cross-functional team promoting AI usage across the company).

For the fifth volume of this series, we spoke with @Kyohei, who is responsible for growth strategies for Mercard and Mercari Mobile in our Fintech Growth Marketing organization, while also promoting the AI transformation of marketing operations as a member of the AI Task Force. We asked him about both what AI adoption looks like on the ground within his organization and his strategies for leveraging AI on a personal level. We hope this article will be helpful for anyone who’s interested in AI but isn’t sure how to get started!

First, have AI execute the tasks while humans focus on decision-making and fine-tuning

@Yurino: Thank you for joining us today, @Kyohei! As both a marketer and a member of the AI Task Force, could you share what your role entails within the task force?

@Kyohei: I break down the marketing workflow and create tools to streamline it with AI. I also build mechanisms to deliver these tools to the team for use.

@Yurino: I see. Is there a particular approach that you keep in mind when promoting AI within the team?

@Kyohei: I believe the key is to first try giving the work to AI and then go from there. Until recently, it was the norm to create documents and proposals manually, but now we first let AI create them and focus our efforts on making decisions and fine-tuning the results. But another thing to remember is that you need to be careful to avoid making AI usage the goal. The important thing is the output; AI is a means to that end.

By keeping those two points in mind, I believe the speed and quality of work can change significantly.

@Yurino: Handing over tasks you’ve always done yourself to AI really does require a shift in mindset. In practice, what steps is the task force taking to encourage team members to try giving the work to AI first?

@Kyohei: The task force creates Notion AI tools that anyone on the team can easily use. We are also introducing command-line tools like Claude Code, but we prioritize working within everyday tools first.

Essentially, we set up pages in Notion with prompts optimized for each type of task—similar to “Skills” in Claude—so that AI can broadly support the work needed for various initiatives.

It’s very simple to use—just mention the page in Notion, and the prompts written on that page will be triggered to support your task. For example, there are prompts to generate specifications, work breakdown structures (WBSs), landing pages, and more from a proposal. All you have to do is give instructions to the AI within Notion, which we’re all already using, so there’s no need to learn any new software.

@Yurino: It’s convenient that you don’t have to leave Notion. Does it give high-quality results right from the start?

@Kyohei: No, but we aren’t aiming for immediate perfection. We focus on having AI execute the tasks first, then having humans make the decisions and fine-tune the results, and work to make that process as easy as possible. Humans review and refine the drafts produced by AI. I believe this approach will make it dramatically faster to produce output compared to having humans write everything from scratch.

Successful AI adoption in teams hinges on a ‘user-first’ design that ensures people will use it

@Yurino: What do you think is important for incorporating AI into marketing operations and achieving successful adoption throughout the organization?

@Kyohei: There are two main points. The first is making sure people actually want to use it—that they know about the tool, understand how to use it, and feel that it’s meaningful. The other is making sure it can be used immediately. I believe that the adoption rate of a tool increases when people can start using it right away, without needing any special preparation.

@Yurino: What led you to those conclusions? I’m also involved in managing the task force, and I sometimes see cases where good tools are not being used much.

@Kyohei: I started working as a member of the task force in January. I had previously tried to introduce an AI tool mainly aimed at developers that operates via the command line, but unfortunately, the usage rate among marketers was not as high as I had hoped.

@Yurino: What do you think was the reason for the low usage rate?

@Kyohei: There were three main reasons. First, non-engineers weren’t familiar with using the command line, making the tool intimidating. Second, the initial setup was difficult. And above all, marketers were constantly working to track KPIs, and the volume of their daily tasks was high, making it difficult to secure time to try new tools. That experience made me realize that being able to pick up and use a tool instantly is actually more important than having a ton of high-end features. So now, our approach is to make things as simple as possible, and if it becomes complicated, we look for another way. Instead of introducing new tools, we decided to integrate AI into the existing tools that everyone uses in their daily work.

@Yurino: What kinds of approaches do you take to simplify things?

@Kyohei: First, one of our selection criteria for tools is whether it can be used in a tool that employees use from their first day. We avoid tools that require lengthy installation or setup, and instead prioritize what we can accomplish using tools everyone is already familiar with.

Speeding up the process: Faster draft creation leading to higher-quality initiatives

@Yurino: What goals does the AI Task Force have in mind for optimizing marketing operations?

@Kyohei: Recently, we have been experimenting with ways to compress the time from planning to draft completion so it can be done within one day. For example, when given a proposal, we aim to be able to output specifications for implementation, LP designs, banner images, and, if necessary, even a functioning page within one day.

Currently, it can take anywhere from several days to several weeks to go from planning an initiative to completing the output. If we can shorten that to one day, we’ll be able to conduct multiple reviews while checking the draft based on actual materials, which will improve the quality of the initiative. Plus, since the hours of work required per initiative will be reduced, we can increase the number of initiatives themselves, ultimately leading to the growth of the entire business.

@Yurino: I see. How specifically will you use AI for this?

@Kyohei: There are two main uses: creation and review. In creation, we use AI to broaden ideas based on past data and to establish the objectives, background, KPIs, and target audience for each initiative. In review, AI might, for example, check for risks based on relevant laws, examine potential fraud perspectives by looking into what kind of fraud might be possible, and verify whether the content resonates from the perspective of the created persona (the expected user).

@Yurino: I’m surprised to hear that AI even checks for fraud perspectives. By having AI handle both creation and review, the process from planning to release must be significantly faster.

@Kyohei: That’s right. In the task force, we have a full set of tools for AI to cover each step of marketing operations. For example:

• Creation of approval documents and proposals
• Automatic generation of specifications and WBSs (schedules)
• Creation of communication plans and LP structure proposals
• Legal checks and secondary reviews from the perspective of customer service

In other words, the entire workflow from planning to the final check before release is supported by AI.

@Yurino: Is there anything else the task force is working on besides the Notion AI tools?

@Kyohei: We’re also concurrently developing automated promotion analysis tools, automated submission tools for distribution settings and landing pages, and tools for mass-producing advertising banners. Right now, we’re in the phase of finalizing the tools, and in the next step, we plan to significantly increase the usage rate across the entire team. Ultimately, we aim for a world where these tools are autonomously connected and executed by AI agents.

Trying things out in his hobbies and applying them to his main job

@Yurino: Do you use AI outside of work as well?

@Kyohei: I use it quite a lot, actually. I run several small businesses as a hobby, such as managing a studio and running a candy apple stand at events, and I use AI in almost every step of the process—from brainstorming ideas to project management. For example, in managing the studio, I built my own AI-powered application for interior coordination and actively use it.

@Yurino: You even go so far as to create apps with AI! It feels like you treat AI more as a co-manager than just a consultant.

@Kyohei: That’s exactly right. I try giving the work to AI first in my hobbies, and when something works well, I apply what I learn to my main job. Recently, whenever I start something new, it’s become second nature to think, “Let’s have AI try it first.”

Starting small first: What do you want to achieve?

@Yurino: What kind of world is the AI Task Force envisioning beyond this?

@Kyohei: The goal is for entire teams to fully utilize AI and reduce the time spent on manual tasks. That way, the freed-up time can be dedicated to refinement and decision-making. The goal we have set for the task force is to ensure that the tools currently under development are adopted by all teams.

@Yurino: What does it mean to you to be AI-Native?

@Kyohei: I believe that someone who is AI-Native is someone who can create output together with AI in line with their objectives. For example, they can share the workload of proposals, designs, and reviews with AI, achieving high quality even by themselves. In the future, this kind of workstyle will become the norm, and I believe the boundaries between what is considered a marketer’s job and a designer’s job will become increasingly blurred.

@Yurino: Conversely, when AI becomes commonplace, what do you think will be expected of individuals?

@Kyohei: In an era where anyone can achieve output using AI, I believe the most important quality will be having a clear vision of what you want to accomplish. As long as you have the curiosity to embrace change, AI will empower you to bring your ideas to life.

@Yurino: If someone reading this article were to start using AI today, where should they begin?

@Kyohei: There’s absolutely no need to use complicated tools right away. What really matters is considering whether using AI will help you reach your goals more quickly or enhance the quality of what you create. It’s essential to think from that starting point.

Also, it’s important to set aside some time for trial and error—even if it’s just a short interlude. Approaching AI with the mindset of having to master it perfectly can often make it hard to carve out the necessary time or take that first step. It’s perfectly fine to start with a casual mindset of “let’s try leaving it to AI first.” From there, you’re sure to arrive at the point where you feel it’s convenient.

What do you want to achieve by using AI? I believe that asking that question is the best starting point for utilizing AI. In the AI Task Force, we aim to change marketers’ workstyles and will continue updating Mercari’s work environment.

@Yurino: Thank you very much!

Check out other articles from this series!