AI検索ドック
: How to use it, features, and the business problems it solves

What is AI検索ドック?

AI検索ドック (AI Search Dock) is a one-off diagnostic service that examines how your company appears in AI answers, provided by Stellagent, Inc. It repeatedly asks ChatGPT, Gemini and Copilot the questions your customers are likely to ask, tallies how often your company and your competitors appear in the answers, and delivers the results as a diagnostic report and a list of actions.

It is not a measurement tool (SaaS) that you keep paying for every month. It is a diagnostic-style service: one product, service or brand is measured in a single batch, and you receive the results. According to the official note post, the service launched on October 1, 2026, and all observations are made in Japanese.

Not a tool you check monthly, but a one-time measurement before you start optimizing

In LLMO/GEO (AI search optimization), the mainstream offering is a subscription monitoring tool where you register prompts and track daily or weekly trends. AI検索ドック plays a different role: it focuses on investigating, in one go and in depth, how AI sees you right now.

One of the customer testimonials on the official site says: "I also appreciate that I can order it once when I need it, rather than signing a monthly contract." The service is designed for situations such as wanting to show the current state in numbers before getting budget approval, or not yet having anyone in-house who can run AI search measurement.

Each question is asked 10 times per AI

AI answers change every time you ask, even for the same question. If you ask ChatGPT once and your company does not appear, you cannot tell whether it "never appears" or "appears 3 times out of 10." AI検索ドック repeats the same question 10 times per AI so that this variation can be counted as a proportion (the recall rate).

The recall rate is "the number of answers in which the name appeared ÷ the number of answers," regardless of whether it was mentioned first. Because a single answer can name several brands, the total across all companies does not add up to 100%. Rankings are ordered by the average recall rate across the three AIs.

What the diagnostic report shows

The official site describes the contents of the report in three parts. Based on the idea that "ranking numbers alone don't tell you what to do next," the service records not just the comparison results but also the answers and the search process behind them.

1. Comparison of your company and competitors

For each question and each AI, the report lists the proportion of answers in which your company and your competitors appeared. You can see at a glance which questions you show up for and which ones you are ceding to competitors.

Source: report sample on the AI検索ドック official site (all company names and figures are fictional; the same applies to the images below)

2. Full answers and cited sources

The answers underlying the figures are kept in full, and the sources cited in each answer can be traced. Once you know which pages the AI relied on, you can identify which of your own pages to fix and which comparison articles to approach about being listed. The official blog explains that, in addition to the report (PDF), a list of all answers and the cited-source data are provided as CSV.

3. Search queries generated by the AI

ChatGPT and similar AIs rephrase a question into search terms, search the web and then answer. AI検索ドック captures those search terms and shows what the AI is prioritizing — use case, price, comparison criteria and so on. In the official sample, the question "What running shoes do you recommend for beginners?" is broken down into search terms such as "running shoes beginner recommended" and "cushioning running shoes comparison."

Report structure

The sample report excerpted on the official site has three parts: "Results," "Interpreting the results" and "Strategy and improvements."

The "Industry map" in particular plots each company on two axes: branded search volume (a proxy for market presence) and AI recall rate. It shows on a single page gaps such as being well known in the market but lacking presence in AI.

How to use

1. (Optional) Try one question with the free diagnosis

   Enter the URL of your company's public page into the free diagnosis on the official site; it drafts candidate questions and, in about 10 minutes, shows the results of asking ChatGPT and Gemini 10 times each. This gives you a rough idea of how often you appear before ordering the full diagnosis.

2. Ask questions in the chat and book a meeting

   In the chat on the official site (AI検索ドック "Miru"), you can ask about price, delivery time, report contents and more, and schedule a consultation with a staff member directly in the chat.

3. Consultation and question design

   Tell them the product to be examined, the competitors you want to compare against, and what you want to know; the provider then designs the 30 questions. You don't need to come up with the questions yourself.

4. Payment

   Payment is made in advance, in a lump sum, by bank transfer (invoice) or credit card. Measurement begins once payment is confirmed.

5. Measurement and recording, analysis and action planning

   Each AI is asked repeatedly, and the full answers, cited sources and search queries are recorded. Your company is then compared with competitors, and concrete actions and priorities are compiled.

6. Receiving the report and the explanation

   The diagnostic report and action list arrive in about 10 business days (one week at the earliest), and you receive an explanation of the contents.

Features

Because AI検索ドック is a service rather than a tool, this section summarizes the work included in the price and the deliverables.

Measure a single question first with the free diagnosis

Separately from the full diagnosis, a free, simplified version — the free diagnosis — is available. Enter a URL, and a market scan (about 1 minute) drafts three candidate questions; the results of asking the chosen question to the AIs 10 times each appear in 5–10 minutes. You can also type your own question, but company names and product names cannot be entered.

Because the free diagnosis covers only one question, it cannot show which questions you appear for and where you lose out to competitors. Seeing the full picture is what the paid diagnosis is for.

Pricing

• The fee applies per product, service or brand. To diagnose several products, you order one diagnosis for each
• There is no monthly fee or minimum contract period

Pros and cons

Pros

• You can order just once, with no monthly contract: easy to budget for when you only want to know the current state before committing to an annual tool contract
• Each question is asked 10 times, so variation shows up as a proportion: the "sometimes it appears, sometimes it doesn't" that one query can't reveal is counted as a recall rate
• Copilot is included as standard: alongside ChatGPT and Gemini, you can measure how you appear in Microsoft Copilot in the same engagement
• You get more than numbers: full answers, cited sources and the AI's search queries are recorded, and the action list is prioritized, making it easy to move on to the next step
• Question design is handled for you: you don't have to figure out which questions to measure with
• You can try it first with the free diagnosis: one question is free, with results in about 10 minutes

Cons and caveats

• A single diagnosis doesn't show trends: to track the effect of your measures over time, you need to order another diagnosis or use a subscription monitoring tool alongside it
• Three AIs are covered as standard: Google AI Mode and AI Overviews are options (¥100,000 each, excl. tax). Perplexity and Claude are not listed as covered
• Payment is in advance, and cancellations and refunds are not possible once the application is confirmed: it helps to settle the product, competitors and questions you want answered internally before ordering
• The price covers recommendations: execution work such as revising pages or producing content is not listed among the items included in the price

Reviews and reputation

AI検索ドック only launched in October 2026, and there are no reviews yet on third-party review sites or social media. The official site carries comments from two people who received a diagnostic report, published under their real names.

Source: "Customer voices" on the AI検索ドック official site (translated by AI BEST SEARCH)

Both comments praise two things: the convenience of ordering a one-off engagement, and an analysis that shows where you stand in your industry. The former corresponds to the pricing model, the latter to the report's "Industry map."

Provider: Stellagent, Inc.

Stellagent, Inc., which operates AI検索ドック, was founded in February 2022. It is based in Minatomirai, Nishi-ku, Yokohama, Kanagawa, and its CEO is Akihiro Suzuki. Its stated business is "developing data connectivity and transaction infrastructure for AI commerce," and it is building infrastructure for an era in which AI understands products and data and takes part in transactions.

In July 2026 the company published a series of surveys on how major AIs recommend home appliances and hotels, and in the same month launched "Stella LLMO," an LLM optimization service specialized for retail and e-commerce. AI検索ドック can be seen as a service that makes this method of collecting and interpreting large numbers of AI answers available to any industry as a single diagnosis. The official blog explains, in Japanese, the basics of AIO, GEO and LLMO, cost benchmarks, comparisons of diagnostic services and more.

Frequently asked questions (FAQ)

How it differs from other LLMO/GEO tools

Most services that measure how you appear in AI search are tools that measure continuously on a monthly subscription. AI検索ドック is a per-engagement diagnosis, which is a different form of delivery.

• The biggest difference is whether you measure once in depth or track continuously. Monthly tools suit tracking trends and checking the effect of measures; AI検索ドック suits understanding the current state and setting priorities before you start
• The idea of repeating the same question to measure the probability of appearing can also be seen in DolphinX AIO's on-demand measurement (asking the same question 10 times in a row). AI検索ドック bundles this at the scale of 30 questions and three AIs into a single engagement, with analysis and recommendations finished by people
• While many monthly tools cover Perplexity and Claude, AI検索ドック includes Microsoft Copilot as standard. The scope you should measure depends on which AIs your customers mainly use
• If you want to keep measuring the issues revealed by the diagnosis afterwards, one option is to combine it with tools you can start on a free plan, such as LLMOinsight or AIsearchmap

The information on this page is based on what was published on the official site as of October 2026. Please check the official site for the latest specifications and pricing.