When candidates search with AI, how should companies make themselves findable?

But is that all students want to know? “Can I move into development with a humanities degree?” “How is placement decided after joining?” “What skills will I learn in that job?” Today some students put questions like these not to a company’s recruiter, but to AI.

In a survey of students graduating in 2027 that Mynavi published in May 2026, 84.9% had used AI in their job search, and 47.6% had asked AI for advice about it. AI is becoming not only a tool for polishing applications, but someone to talk to about where to work.¹

Companies need to adapt. Not by stopping students from using AI, but by building an environment where students, and their AI, can learn about a company accurately, compare it and ask questions. Here is what that takes.

Do not leave your hiring information to the job boards

Students have usually started their search on the big job portals and at joint company fairs. Those will stay important. But that does not mean a company’s own careers site can do nothing more than point to the portals.

Say a student asks AI: “Is there a company in Tokyo where I can learn data analysis from scratch? I’d also like to know how often placement requests are honoured.”

If all there is to go on is a short company profile on a portal, AI has little to draw on. If the company’s own site publishes, for each role, the work, the skills needed, location, how placement is decided, training, and pay and allowances, candidates can decide from that. AI can organise whatever it was able to read.

So start by reviewing your careers site’s SEO. Give each role a job page with its own URL. Write the work and conditions in the body text, not only inside images or PDFs. Link pages sensibly so search engines can reach the content. Do not leave pages for closed roles lying around.

Job pages can also use Google’s JobPosting structured data. It tells Google’s job search about a listing so the listing can be eligible for that search experience; it does not guarantee it will appear.²

I think SEO and readiness for LLMs will become not a special growth tactic, but the basic hiring infrastructure every company needs. This is not about quitting the portals. It is about keeping both the external channels and information you are responsible for and can update yourself.

Let anyone take “information for AI” with one click

What if students visiting a careers site could do this?

• Copy this page as Markdown: copy the work and requirements with headings and tables intact.
• Copy all the hiring information: bundle the related public pages, such as job descriptions, pay, placement, training and internships, into one.
• Ask your own AI: paste what you copied into ChatGPT, Claude or another assistant, and ask “does this company fit what I want?” or “what other conditions should I check?”

The point is not to make the company’s own chatbot the only option.

Students already have an AI they are used to. It may already know what they want and what they have asked before. Handing official information to their own AI beats repeating the same explanation to a different AI at every company.

Markdown suits this. It carries headings, lists, links and tables without the extra visual chrome. In February 2026 Cloudflare also announced a feature that returns a site’s HTML as Markdown on request.³

However, Markdown will not raise your ranking in AI search. Google states clearly that special AI files and Markdown are not required to appear in its AI search features.⁴ This proposal is not a ranking trick. It is about delivering official hiring information in a form that is easy for both people and AI to use.

When copying, include the source URL, the date updated and the role it covers, so that old conditions are not treated as current forever once a posting changes. And make clear, before anything goes to AI, that only public information is copied. Students’ personal information, such as their application history or questions they asked, must never be included by accident.

Give each company an AI recruiter

Making information easier to find does not settle everything.

“The posting says science graduates are welcome. Can I still apply with a humanities degree?”

“I’m interested in working abroad. Can I ask for that in my first placement?”

“How many hours a week does the long-term internship take?”

Students’ questions do not fit on one page. So each company sets up an official AI recruiter: one place to ask, drawing on all the public information across the careers site.

This is not a chatbot that lists frequently asked questions. It answers questions that cut across roles and conditions, points to the relevant pages when needed, and when it does not know, says “this needs checking” and hands over to someone in HR. For a humanities student who wants a development role, for example, it can walk through eligibility, the basics needed, what to study beforehand and which internships they can join, all from public information.

Draw the line on what AI must not do, though. It does not guess at non-public selection criteria. It does not promise an offer or a placement. Every answer cites its source and date. Complex questions and objections to an answer go to a person.

More important still is keeping these questions apart from assessment. If asking “Is there a lot of overtime here?” or “Will my background count against me?” can end up in an evaluation, students will not ask openly. Be clear about how long question logs are kept, what they are used for and how to delete them, and make it a principle that they are not used in selection.

An AI recruiter is not there to screen candidates out. It is a place where students can think through, with confidence, whether to work at that company.

Do not tell only the flattering parts

However much you prepare for AI, it is pointless if the source information is vague.

If you write “lots of autonomy”, what exactly can people decide for themselves? If you write “thorough training”, how long is it, what does it cover and who runs it? If you write “placement takes preferences into account”, how is it finally decided?

Write about the difficulties and constraints, not only the appeal: busy seasons, the chance of relocation, the load of dealing with customers, the possibility of doing work you did not ask for. Hiding conditions to win more hires is not good hiring for either side if expectations and reality turn out far apart.

In the previous post I looked at hiring where companies and students check each other through long-term internships. The careers site, as the way in, needs the same thinking. Which team, doing what work, learning what, and leading straight into hiring for which role. Only once that is said can students judge whether taking part is worth it.

In a Mynavi survey from February 2026, 85.6% of students graduating in 2027 had taken part in internships or other career development programmes.⁵ That figure is not for long-term internships alone. I read it as a sign that, as companies create all kinds of touchpoints, the chance to understand the work beforehand matters more to students.

A future where students’ AI and companies’ AI talk to each other

From here on, this is a prediction.

Instead of researching companies one by one, students may increasingly ask their own AI: “Compare companies that match where I want to work, the role, the pay and what I want to learn.” Their AI fetches each company’s public information and asks the official AI recruiter about what is missing. People then look at the result and choose the companies they actually want to meet.

Seen this way, the first thing a company should do is not build a flashy AI recruiter. It is to make sure its hiring information can be retrieved accurately.

With a sound system for public job information, the same information serves a student browsing, a copy to their own AI, the company’s official AI answering, and, in future, AI asking AI. If the information is inconsistent, every new way in only spreads the errors further.

Of course, I do not think a world where only AI-recommended companies get chosen is automatically a good thing. Large companies with plenty of information could gain the most, and AI judgement could narrow students’ options. That is exactly why candidates should still be able to read the original, change the basis of comparison and ask directly.

Where to start

There is a lot you can do without installing a large AI system.

First, sort out the hiring information. Put a job page for each role on your own site and keep it current. Organise the work, requirements, pay, location, placement, development, and how internships connect to hiring. Cover the SEO basics and the right structured data.

Next, let students use that information. Add a Markdown copy to each page and a way to take related pages together. Where needed, serve Markdown so AI can read it too. Generate everything from one source so updates never drift apart.

Then add an official AI recruiter. Start by guiding people accurately through public information, and pass questions it cannot answer to a person. Learn what students are asking, without identifying anyone, and fill the gaps in your hiring pages.

To measure the effect, page views and chat counts are not enough. Check whether students’ questions were answered, whether misunderstandings about conditions after applying went down, and whether gaps in understanding at internships and interviews shrank.

A careers site does not have to stop at advertising that makes the company look good. It can be where students find work that suits them, and where companies describe that work accurately.

In the age of AI, the goal is not to be a company AI likes. It is to be a company whose reality comes through even when researched with AI, so people can choose it with confidence.

This post is a proposal based on public information as of 14 October 2026. Future possibilities, such as students’ AI querying hiring information from companies, are described as predictions. Translated from the Japanese original.

• #Hiring
• #Generative AI
• #Careers sites
• #SEO