AI Hack Inc. (Headquarters: Minato-ku, Tokyo; CEO: Daisuke Nakamichi; hereinafter “AI Hack”) is pleased to announce its “Ground Truth Data Platform” concept, under which a “ground truth database” for each company is maintained as a vector database and a range of capabilities is delivered through MCP (Model Context Protocol). As the first step under this concept, the company has begun providing a hallucination detection feature in its AIO (AI Optimization) analytics tool, “AI Hack,” to a limited number of partner companies ahead of wider availability, and has also launched “AI Hack Agentify,” an agentic web tool. AI Hack has additionally begun “AIO Media Partners,” a new initiative that supports the citation of corporate information by generative AI through co-creation with third-party media.
Background and Purpose
As generative AI services such as ChatGPT, Gemini and Google AI Overviews become the primary point of contact for information, “zero-click” behavior—in which users reach decisions from AI answers alone, without clicking through to search results—continues to spread. For companies, whether they are correctly recognized and cited within AI answers now determines the entry point to their business.
At the same time, AI answers always carry the risk of containing factually incorrect statements (hallucinations) or outdated information. Companies will also need to prepare for the “agentic web,” in which AI agents visit websites on behalf of users to obtain information and complete procedures.
To address these challenges, AI Hack is advancing the Ground Truth Data Platform concept as a platform for running the entire cycle of measure → correct → supply the information to AI → re-measure. The company organizes a single, company-specific “ground truth database,” composed solely of that company's accurate facts, as a vector database, and uses MCP as a gateway to deliver measurement, detection and supply capabilities on one shared platform.
About the “Ground Truth Data Platform” Concept
The platform consists of two layers: a base layer and an application layer.
- Base Layer (Core)
The “ground truth database,” organized as a customer-dedicated vector database, together with an MCP gateway responsible for authentication, authorization, and auditing. It holds and supplies only those facts a company has substantiated, in a form that AI engines and AI agents can reference.
- Application Layer
The set of features that run on top of the base layer. Three features are currently available: AIO analysis, hallucination detection, and AI Hack Agentify. AI Hack will continue to expand the lineup on the same platform, including article writing and rewriting, automatic hallucination correction, and support for agentic commerce.
New Feature: Hallucination Detection Now Available in the AIO Analytics Tool “AI Hack”
This feature detects factually incorrect statements and outdated information in what AI says about a company. Using the ground truth database as the reference, it compares AI answers against those facts and reports errors, outdated content and omissions. Through a loop of detection → notification → correction → re-measurement, it helps companies maintain the accuracy of AI answers on an ongoing basis, supporting the early identification and remediation of brand damage risk. At launch, this feature is provided to a limited number of partner companies ahead of wider availability. AI Hack plans to expand the scope of provision in stages.
Joint Proposals with “AI Risk Checker,” an AI Risk Assessment Tool Provided by Priv Tech
For hallucination detection to work in practice, the design of verification prompts—what to ask about one's own company in order to bring risks to light—is essential. For this step, proposals are available in combination with “AI Risk Checker,” an AI risk assessment tool provided by group company Priv Tech Inc. (Headquarters: Minato-ku, Tokyo; CEO: Daisuke Nakamichi).
By loading the websites, catalogs, manuals and other materials of a company, service or product into AI Risk Checker, verification prompts specific to that company or product are generated. Registering those prompts in AI Hack and comparing the resulting AI answers against the ground truth database enables hallucination detection aligned with the issues specific to each company.
New Service: Launch of “AI Hack Agentify,” an Agentic Web Tool
“AI Hack Agentify” is an agentic web feature that allows websites to be operated through natural language. By installing a single line of code, both users and AI agents can drive a website with words—from search and guidance through to execution (NLWeb / MCP compatible).
It evolves conventional keyword-matching site search into a conversational experience that understands intent, answers and acts, preparing customer touchpoints for the coming era of AI agents. Widget colors, icons and initial messages can be customized from the administrator page, and multiple domains can be managed centrally.
About the AIO Analytics Tool “AI Hack”
“AI Hack” is an AIO analytics tool that submits prompts to multiple generative AI models (such as ChatGPT) to visualize and score their responses from multiple perspectives.
- Visualization of AI-Generated Responses
By registering prompts (questions), responses are retrieved from multiple AI/LLMs and the resulting AI-generated content is visualized.
- Multi-Dimensional Analysis Using Five AIO Scores
AI-generated responses are evaluated using multiple metrics, including Priority Score, Positive Score, Negative Score, Mention Volume Score, and Domain Citation Score, enabling comprehensive analysis.
- Competitive Analysis
In addition to information about the company itself, information related to multiple competing companies or services is simultaneously scored to facilitate competitive benchmarking.
Supported AI engines span four major providers: OpenAI (ChatGPT), Google (Gemini, AI Overviews, AI Mode), Perplexity, and Anthropic (Claude).
New Initiative: AIO Media Partners
AI Hack has also begun “AIO Media Partners,” an initiative that supports the citation of corporate information by generative AI through co-creation with third-party media. When generating answers, generative AI places weight not only on a company's official website but also on “mentions outside the company,” such as specialist media and comparison or review articles. Through this initiative, AI Hack will co-create exposure in third-party media that AI cites, together with its media partners, on two fronts. The first is the joint operation of a specialist publication as an original journal under the Business Journal brand, a business media outlet run by Angle Create Inc. (the authority layer). The second is article distribution across more than 1,000 domains through the networks of affiliate partners, including COLOR ADS Inc.
Corporate Website Renewal
In line with this announcement, AI Hack has fully renewed its corporate website (https://ai-hack.co.jp/). Alongside an overview of the Ground Truth Data Platform concept, the renewed site introduces each of the company's services: the AIO analytics tool “AI Hack,” AIO consulting, the agentic web tool “AI Hack Agentify,” development support for Apps in ChatGPT, and AIO Media Partners.
Future Plans
Beyond measuring and improving AIO, AI Hack will expand this platform as the foundation that supplies corporate ground truth data to AI, supporting end-to-end business optimization for the era of the “agentic web and commerce,” in which AI agents converse and transact.
About AI Hack
AI Hack, Inc. provides analytics-driven tools and services that remove barriers to generative AI adoption—helping clients accelerate digital transformation while mitigating privacy, copyright, and other compliance risks. The company specializes in AIO (AI Optimization), a structured framework for visualizing and optimizing AI search responses (also known as GEO or LLMO), enabling businesses to track AI behavior, improve brand alignment, and ensure accurate, ethical AI outputs.