Building “the AI that finds the next AI.”

FOUNDER'S BLOG

Building “AI that finds the next AI.”

Reducing technical due diligence to 30 seconds with AI. Announcing the establishment of LINEdot. Inc.

— On the Establishment of LINEdot. Inc.

May 1, 2026 Masato Furuno, Representative Director

Today, we established LINEdot. Inc. We are a startup that automates technical due diligence using AI. As the founder, I’d like to speak as candidly as possible about why I founded this company and what we aim to achieve.

BACKGROUND

There are questions that “search” alone cannot answer

“Which company will be the next big thing in this technology sector?” “How will the ecosystem change if Company A acquires Company B?”—If you’re an investor or work for a business, you likely grapple with questions like these on a daily basis. But to be honest, it’s difficult to answer these questions using current search technology or RAG.

Search is good at “finding information that already exists.” But what business decision-making truly requires is not the “discovery” of information, but “understanding” and “reasoning.” Relationships between companies, the causal structures of technology trends, the ripple effects of M&A—these are things that search, by its very nature, cannot deduce.

We want to bridge this gap. That is why we launched LINEdot.

PRODUCT

“WARP DD” — Our First Product

Our first product, “WARP DD,” is a platform that uses AI to automate technical due diligence for global startups. In a nutshell, it is “an AI that searches for and evaluates technology companies around the world simply by entering a keyword.”

Automatically searches for and scores global tech companies with just a keyword (reports generated in about 30 seconds on average)

Covers major global tech data sources (research papers, patents, GitHub, Crunchbase, etc.) and integrates deep analysis using multi-LLM

Six-axis quantitative evaluation and de facto trajectory prediction using a proprietary three-layer AI analysis engine

Automatic generation of due diligence reports (PDF/PPTX) including executive summaries, detailed analysis, and recommendations

It reduces the time required for technology due diligence—which previously took human analysts weeks to months—to an average of about 30 seconds. You might be wondering, “Can it really do that?” In a proof-of-concept (PoC) validation of our core algorithm, we achieved an overall accuracy rate of 89% in a two-year backtest covering 100 companies in the AI/ML sector.

IMPACT

The Impact of WARP DD: By the Numbers

How much of an impact does hybrid inference using our proprietary multi-AI model have compared to conventional technical due diligence? We’ll show you with concrete figures.

Evaluation Item	Conventional Technology DD	WARP DD（AI）
Time	Weeks to months	Approximately 30 seconds on average
Cost	High (Hiring external experts)	1/10 to 1/100 of conventional costs
Data Coverage	Limited (Scope of data collectable manually)	Major global data sources × Multi-LLM deep analysis
Consistency of Evaluation	Dependent on consultants (Subjective judgment)	Reproducible AI-driven 6-axis quantitative evaluation
Predictive Capabilities	None (Current analysis only)	Future potential assessment using de facto trajectory prediction AI
Backtest Accuracy Rate	N/A	89% (Verified over 2 years across 100 AI/ML companies)

* Backtesting: Predicted and verified the trajectory two years from a June 2023 snapshot for 100 companies in the AI/ML sector.

* Cost comparison is based on a comparison with the fees for commissioning technical due diligence from external experts.

CORE TECHNOLOGY

Three AI Models That “Understand” the World

At the heart of WARP DD is a proprietary three-layer AI analysis engine. It analyzes complex technology landscapes—which a single model cannot fully capture—from three distinct perspectives.

MODEL 01 Ecosystem Analysis AI	Based on graph neural networks. Models the relational structure among technology, research, and the market to capture the dynamics of the entire ecosystem.
MODEL 02 Technology Profiling AI	Quantitatively evaluates competitiveness across six dimensions based on academic papers, patents, and technical documents. Objectively visualizes the strengths and weaknesses of target companies and technologies.
MODEL 03 De Facto Trajectory Prediction AI	Calculates the degree of alignment with the growth patterns of de facto industry leaders to predict the likelihood of a technology becoming an industry standard. Also supports “what-if” analysis.。

These three models work together to perform in-depth analysis of major global technical data sources using a multi-LLM system. In a two-year backtest, the system achieved 100% accuracy in predicting high-growth companies, 92% for stable-growth companies, and 75% for detecting M&A signals. It successfully identified the emergence of high-growth companies and major M&A deals in advance.

VISION

What Lies Ahead—The Technology World Model

“WARP DD” is merely a starting point. Our true goal is the “Technology World Model”—a next-generation technology intelligence platform that autonomously builds and updates an internal model of the technology landscape, supporting decision-making through “understanding and inference” rather than mere “search.”

In fact, our current three-layer AI engine already implicitly incorporates the four-layer structure of this world model. The Ecosystem Analysis AI corresponds to the perception layer and the world state layer, the Technology Profiling AI corresponds to the inference layer, and the De Facto Trajectory Prediction AI corresponds to counterfactual simulation. Our roadmap to 2030 involves making these components autonomous and scaling them up.

ARCHITECTURE

Four-Layer Autonomous Intelligence Architecture

LAYER 01 Perception	Perception Layer — Continuously streams and collects raw data on the technological world from distributed data sources. Rather than batch searches, it continuously observes changes in the world in real time.
LAYER 02 World State	World State Layer — Maintains the relational structure of companies, technology, markets, and investments using a hybrid of dynamic graphs and continuous embedding spaces. Automatically restructures itself based on new information.
LAYER 03 Reasoning	Reasoning Layer — Executes multi-step reasoning, counterfactual reasoning, and predictive reasoning on the World Model. Understands causality rather than mere correlation.
LAYER 04 Action	Action Layer — Generates DD reports, issues risk alerts, and automatically recommends PoC candidates. Takes autonomous actions based on reasoning results.

ROADMAP

WARP DD → Technology World Model

✓ Completed 2026 WARP DD Hybrid 3-Model Integration Cross-domain Search & Automated Reporting

2027-28 Knowledge Graph Dynamic KG Construction + Structural Reasoning Model Scaling

2029-30 Technology World Model Fully Autonomous World Model. Counterfactual Reasoning, Prediction, and Causal Discovery

OUR BELIEF

What We Believe

The world of technology is an ecosystem where countless players are intricately intertwined. To truly “understand” it, three key capabilities are essential: visualizing relationship structures as graphs, predicting changes over time, and inferring causal relationships.

By developing and integrating our proprietary AI models, each specialized for these areas, we are creating a world where AI can answer “questions that search engines couldn’t answer.”

LINEdot. was created to build that future.

COMPANY INFO

Company	LINEdot. Inc.
CEO	Masato Furuno
Date of Incorporation	2026.5.1
Address	Aoyama Marutake building 6F 3-1-36 Minami Aoyama Minato-ku, Tokyo
Business Activities	Development and operation of an AI-powered due diligence automation platform
URL	https://www.linedotjp.com/
Email	hello@linedotjp.com