"PR for AI" Comes After SEO: AI Hack's Daisuke Nakamichi on How to Approach AIO

Nakamichi's position is clear. The rules of web marketing have already changed. The focus is shifting from building out your own website to how your company is talked about in external, third-party media. That, he argues, is why companies now need to do PR aimed at AI.

According to Nakamichi, AIO stands for AI Optimization. Where traditional SEO (Search Engine Optimization) optimized for search engines, AIO refers to efforts to optimize how AI answers questions, now that people use AI to look things up, along with the whole surrounding field. The same field is also called LLMO (LLM Optimization) or GEO, but AI Hack uses "AIO."

The reason for the choice was practical. Nakamichi was initially torn between LLMO and AIO, but went with AIO because LLMO is a tongue-twister when you say it out loud. GEO overlaps with a corporate group of the same name, so searching for "GEO" won't get you to the top of the results. Nakamichi's view is that the field is now mostly called either LLMO or AIO.

AI Hack was founded in July 2024, when the term AIO barely existed. Inquiries surged starting around summer 2025. Because the company has done almost no marketing, the total number of inquiries isn't large, but many of them come from major corporations.

Iijima said that in 2025 he clearly felt traffic from Google drop on his own web media outlet. He confessed that he sees AI Overviews, and users who do their research with AI, as a threat.

Nakamichi replied that this is happening across the entire industry. An AI answer (the AI-generated summary, or AI Overview) appears at the top of Google's search results, and the user's research ends there. The links listed below go unclicked. This phenomenon is called "zero-click search." Nakamichi's view is that even if corporate web managers aren't saying so out loud, there are probably few cases where traffic is going up.

Many companies come to him asking to have their brand or links featured in AI Overviews. But Nakamichi said what he wants to get across is that, as a matter of Google's own strategy, its role of sending users to websites is coming to an end.

What Nakamichi is watching closely is Google's "AI Mode." It goes a step beyond AI Overviews, letting users dig deeper into a topic through a conversation with AI. The current interface is built to steer users from AI Overviews into AI Mode with a prompt to "dive deeper in AI Mode." In Nakamichi's assessment, Google has been assembling features with the aim of having people finish their research within a conversation in AI Mode, and even complete purchases there.

E-commerce businesses, for example, have until now assumed they would drive traffic to their online stores through Google Ads and SEO. If everything is completed within Google's platform, the design of KPIs, and the very objectives themselves, will change. Things need to be assessed based on a new way of thinking, Nakamichi says. When Iijima remarked that "the rules are changing," Nakamichi responded, "It's already a game change."

So should companies rethink their SEO from scratch? Nakamichi believes the behavior pattern of searching, clicking, and going to a website is far from disappearing. At the same time, it's certain that some research is being taken over by AI. That's why, he says, companies need to gradually shift things like budget allocation toward AI.

As advice for small and mid-sized companies and newly launched businesses, Nakamichi positioned SEO as "absolutely necessary as a minimum." Do SEO, and then layer additional measures on top, such as exposure in external media and writing in a style AI favors. The way he frames it, there are simply more things you have to do.

The first difference between SEO and AIO that Nakamichi pointed to was the difficulty of analysis. With SEO, if you search for something like "shoes e-commerce," you can immediately see where your site ranks. With AI, that kind of information isn't disclosed. So the first step, Nakamichi says, is to analyze the current situation, that is, to examine how AI perceives your brand and URLs and how it answers questions about them.

Is there a right answer for AIO at this point? Nakamichi answered no, there isn't, acknowledging that a lot of it is trial and error. With SEO, you basically just had to optimize for Google. Yahoo also uses Google's search engine, so optimizing for Google worked for Yahoo to a certain extent too. AI, however, is in a state of fragmented competition. Within Google there are AI Mode and AI Overviews, and there's Gemini as a standalone app. Beyond that there's OpenAI's ChatGPT, China's DeepSeek, Perplexity, and others, and each gives completely different answers.

There are also differences between models. As Nakamichi recalled, GPT-5.0 came out around August 2025, followed by 5.1 and 5.2, with 5.2 being the latest at the time of recording. Each model produces answers and output a little differently. You have to handle each company's AI, and also keep up with the new models each company keeps releasing. As a result, Nakamichi said, the workload is considerably greater than with SEO analysis.

In the analysis, you look at how prominently your brand appears in an AI's answer, whether it's positive or negative, and whether your own domain is cited. When AI generates an answer, it also shows the URLs it referenced. According to Nakamichi, what stands out there are third-party media. There is a strong tendency for articles and outlets that compare and evaluate the features of Company A's, B's, and C's services from a neutral standpoint to be cited.

In SEO terms, off-site factors become overwhelmingly more important than on-site ones. The key, Nakamichi explains, is how you get exposure on domains other than your own. Iijima likened SEO to "cultivating your own land" by creating content on your own site. AIO, by contrast, requires outward-facing measures. It was here that Nakamichi said, "PR aimed at AI is becoming important."

Why does the outside matter so much? Nakamichi explains that, compared with the old mechanism of crawling based on site structure, AI has become "smart in a sense," gathering information from many places, compiling it, and generating answers. The scope of the work widens and doesn't end with your own site. For that reason, he added, it will likely be a heavy burden for small companies that find it hard to get exposure in external media.

As outlets that tend to be cited, Nakamichi named note, PR TIMES, and Google-owned YouTube. These three come up frequently in analyses with AI Hack's tool as well, he says. Comparison sites also appear very often. It seems they are valued for comparing things from a third-party perspective.

There is a tricky aspect, however. At present, affiliate sites are also frequently cited. Affiliate marketing is fundamentally a performance-based model that pays according to conversions. Getting listed for the sake of AI optimization misses that purpose, and it doesn't generate direct revenue for the affiliate media either. As a result, Nakamichi says he has recently been hearing that some affiliate media have started offering a menu where they get paid for "listings for AI optimization," separate from performance-based fees.

When Iijima asked what makes these outlets strong, Nakamichi said PR TIMES is easy to understand. His basis is his own company's experience. Even for a small company like AI Hack, publishing a single press release on PR TIMES gets it republished by major publishers on their own media. The republished articles state "Source: PR TIMES." From the AI's perspective, PR TIMES articles are being republished and referenced by major outlets, so Nakamichi speculates that this has greatly raised the standing of PR TIMES as the original source.

In addition, because PR TIMES is a venue for PR activity, it has a high degree of public character and gathers the latest information. Nakamichi's view is that this probably makes it convenient for AI as a medium where it can go to get up-to-date information. As for why note is strong, Iijima said he "wanted to dig into that," but no specific explanation was given.

On what information to place in external media, Nakamichi first noted that it depends on the industry and business type, as well as on current trends in AI answers and the results of analysis. With that caveat, he pointed out that one-sidedly declaring "our product is great" tends not to resonate. What's needed, he says, is content that compares options neutrally from a third-party perspective, along with evidence demonstrating the strengths of your product.

For cosmetics, it becomes important to borrow the words of outside experts, such as key opinion leaders. When Iijima checked whether this meant moving away from advertising-style messaging and closer to content made by real people, like UGC (user-generated content), Nakamichi replied that UGC, the user's perspective, and word-of-mouth information are very important.

Iijima, who comes from a PR background, said the more he heard, the more AIO seemed close to publicity work, in other words, public relations. According to Nakamichi, inquiries are in fact coming directly from PR departments. SEO was handled by marketing or web departments, but that picture is changing. Because AI also carries risks such as hallucinations and brand damage, inquiries are coming from risk management departments as well.

Iijima suggested that, just as PR teams sometimes run official corporate social media accounts or the CEO's X account, AIO might fit in surprisingly smoothly if it were tied to PR. In response, Nakamichi pointed out that it will be a cross-functional effort involving the foundations of SEO, brand risk management, and PR objectives, and that collaboration with other departments may be essential. Existing departments alone can't fully handle it, so executives must recognize it as an issue and think about how the company as a whole will tackle it. That is Nakamichi's conclusion.

On the budget side, Nakamichi says "nobody has a budget." No company has yet set aside a budget specifically for AIO. The recording took place in late January, and he says he occasionally gets projects where companies want to first try an analysis using budget remaining before the end of the fiscal year (end of March).

Iijima pointed out that even though ChatGPT and similar tools are so widely used, AI alone remains untouched, a blank spot within marketing. Running corporate TikTok accounts was also at the "how do we even do this?" stage at first, but now many companies are doing it. Iijima's assessment is that AIO seems to be where TikTok was a little while ago. Nakamichi responded that this year it's "probably going to explode." He says AIO consulting firms and a layer of analytics tools like AI Hack's have been emerging both in Japan and overseas.

Iijima feels that many companies are still at the information-gathering stage and haven't yet come to see it as a management issue. Nakamichi, on the other hand, says major companies have already noticed and started moving. In particular, companies whose business models rely on SEO search results as a primary revenue channel have a strong sense of urgency and are taking it seriously.

Nakamichi explained how AI Hack's analytics tool works as follows. Users register in the admin dashboard the questions that people are likely to ask AI. AI Hack sends these, via APIs and other means, to commonly used AIs such as AI Overviews, AI Mode, and Perplexity. It then retrieves the answers and the source URLs. How prominently the client's brand appears in the answer, whether it's positive or negative, how many characters it takes up in the answer, and whether the domain is cited are visualized as an "AIO Score."

Nakamichi said that working on projects is interesting because unexpected results come up, and he gave beer as an example. People would probably think of Kirin, Asahi, Suntory, and Sapporo. But the analysis shows that "Yona Yona (Yona Yona Ale) is strong."

Nakamichi speculates that Yona Yona Ale has very distinctive concepts for each of its brands, such as "Suiyoubi no Neko," and that AI may find it easy to learn those distinctive expressions and concepts. Yet Yona Yona Ale is also a brand that only beer enthusiasts really know. The market share and brand awareness that humans perceive may differ from those that AI perceives, Nakamichi says.

Iijima responded that for people who grew up watching TV commercials in the Heisei era, "beer means Super Dry" has been ingrained in their minds. Nakamichi explained that in the world of AI, it's as if everyone is "starting together from zero." AI physically can't access TV commercials. It learns based on the data and information it can access, and different forces are at work there than the perceptions that have been ingrained in humans. That, Nakamichi argues, is why "PR activities aimed at AI" are necessary. Iijima added that younger generations, who grew up gathering information mainly from the web and social media, may find this easier to grasp. Nakamichi also noted that each AI has its own preferences for how it goes about getting information, and appears to make judgments based on public character and authority. There are techniques that take this into account, he said.

In addition to its AIO tool, AI Hack is also developing an "agentic web." Nakamichi explains it this way. Until now, websites have been optimized for humans to view, and Google's SEO has also rewarded content that is easy for humans to read. In the future, however, each person will have an AI agent, and that agent will go around various sites on their behalf gathering information, and in some cases even make purchases for them. Rather than humans directly operating websites and apps, AI will operate them and gather information. The agentic web, he says, is a concept premised on this worldview.

Asked whether design will become unnecessary, Nakamichi replied that it isn't necessarily fine to drop it entirely, since AI can also learn and recognize design. Even so, he believes the world in which people click links and look at websites one by one will gradually be replaced by AI. Nakamichi thinks human page views will disappear. Since AI agent access can be measured to some extent, new KPIs such as an AI CVR may emerge in the future, he said. When Iijima asked whether there might also be "heatmaps for AI," Nakamichi answered, "There might be."

Nakamichi emphasized that the premise that websites are something humans view, clicking links to drive traffic, has already changed. Unless you revise your approach to SEO on this premise, he says, you won't be able to keep up.

Finally, drawing on more than 20 years of involvement with the web, Nakamichi summed up as follows. The structure of the web industry and digital marketing is changing. First, accept that honestly, and be prepared to shift toward next-generation platforms and forms of customer acquisition. Then, without fear, unravel and address the things that are becoming necessary right in front of you, such as AIO, one by one. That said, he acknowledges that even he doesn't yet have a right answer as to what will reliably work.