A large Machine Learning community in Japan

<p><a href="https://www.mlt.ai/" target="_blank" rel="noopener" aria-label="MLT (opens in a new tab)">MLT</a> is a community and an award-winning nonprofit organization based in Tokyo, Japan. More than 10,000 members from all over the world. Since 2017, organized more than 300 talks, workshops, and study sessions in machine learning, deep learning, reinforcement learning, and math.</p>


large_mlt_transparent_black_font_e1571189935683_50dd637c0a.png

medium_mlt_transparent_black_font_e1571189935683_50dd637c0a.png

small_mlt_transparent_black_font_e1571189935683_50dd637c0a.png

thumbnail_mlt_transparent_black_font_e1571189935683_50dd637c0a.png

mlt_transparent_black_font_e1571189935683_50dd637c0a.png

small_mlt2_a7affc4128.webp

thumbnail_mlt2_a7affc4128.webp

mlt2_a7affc4128.webp

small_mlt4_6a49c502ea.webp

thumbnail_mlt4_6a49c502ea.webp

mlt4_6a49c502ea.webp

small_mlt1_d360bda018.webp

thumbnail_mlt1_d360bda018.webp

mlt1_d360bda018.webp

small_mlt3_dfdf9c06f5.webp

thumbnail_mlt3_dfdf9c06f5.webp

mlt3_dfdf9c06f5.webp

About

Services

Blog

Kabuken

Labs

<noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-MPDT54N"
height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript>

Homepage

AI services, performance and security engineering

nav-card-services.webp

Request information, stay in touch with us

nav-card-contact.webp

Product updates, upcoming events and trending news

nav-card-news.webp

How we built a data platform on Japanese regulatory filings, from XBRL extraction to AI integration

nav-card-case-study-kabuken.webp

We have published a new article on the blog: a step-by-step comparison of four Japanese automakers (Toyota Motor, Honda Motor, Suzuki Motor and SUBARU) built only from their annual securities reports on EDINET and read through Claude with the Kabuken MCP connector.

You get the prompt for each step, the Kabuken tools it calls and the resulting tables: the headline figures, three fiscal years each from its own filing, segment and regional breakdowns and, as an optional step, the large-shareholding reports. Every figure carries the EDINET document ID of the filing it came from, so you can open the filing and check it.

[Read the article](https://intellics.ai/blog/post/19)

Source: EDINET (FSA Japan), PDL 1.0. Processed by Kabuken.

New article: compare a sector from its filings with Claude and Kabuken MCP

An open filing with bar charts under a magnifying glass, and four cars lined up below it.

kabuken-automakers-filings-strip-wide.webp

kabuken-automakers-filing-magnifier-cars-1142w.webp

Kabuken is now listed on mcpservers.org, a web catalog of MCP servers.

MCP is a standard that lets AI apps use outside tools and data. Kabuken is an MCP server for the financials of Japanese listed companies. The figures come from EDINET filings, the disclosure system of Japan's Financial Services Agency. For each figure, Kabuken returns the source filing.

You can look up financial statements, business and geographic segments, major shareholders and the filing list for a company. Labels are in English and Japanese.

The catalog shows an install line for clients such as Claude Code and Cursor:

`npx add-mcp 'https://kabuken.intellics.ai/mcp'`

The server is read-only. It does not provide stock prices or investment advice.

[See the listing on mcpservers.org](https://mcpservers.org/servers/kabuken-intellics-ai) | [Setup guide](https://kabuken.intellics.ai/docs/en/api/connecting/?utm_source=intellics&utm_medium=news&utm_campaign=mcpservers-listing)

Source: EDINET (https://disclosure2.edinet-fsa.go.jp/), PDL1.0 (https://www.digital.go.jp/resources/open_data/public_data_license_v1.0). Processed by Kabuken.

Kabuken is now listed on mcpservers.org

Kabuken's card in a stack of MCP server catalog cards, linked to a filing with a chart

kabuken-mcp-catalog-card-stack-to-filing-tall.webp

kabuken-xbrl-to-mcp-diagram.svg

Kabuken, our platform for the financials of Japanese listed companies, is listed in Anthropic's Claude connector directory. Now it can also draw.

Ask Claude for the income flow diagram of a filing. A Sankey diagram appears in the chat in Claude on the web. Revenue and other income are on the left. Costs and net profit are on the right. If the company reported a net loss, the loss is on the left.

Labels follow the language of your request, English or Japanese. Under the diagram you see the EDINET filing ID, the period, the EDINET source, the PDL 1.0 licence and "Processed by Kabuken".

Kabuken is read-only. It gives no stock prices and no investment advice.

Try this prompt: "Show me an income flow diagram of the latest annual securities report of Toyota Motor."

[See the listing in the Claude directory](https://claude.ai/directory/kabuken) | [Setup guide](https://kabuken.intellics.ai/docs/en/api/connecting/) | [Kabuken](https://kabuken.intellics.ai/?utm_source=intellics_news&utm_medium=referral&utm_campaign=claude_directory_sankey)

Source: EDINET (FSA Japan), PDL 1.0. Processed by Kabuken.

Kabuken now draws a filing's income flow inside Claude

Income flow diagram of Toyota Motor's annual securities report shown inside Claude, with revenue on the left, costs and net profit on the right, and the EDINET source line under it.

kabuken-income-flow-toyota-en-tile-wide.webp

kabuken-income-flow-toyota-en-1567w.webp

Kabuken now ships ready-made skills. A skill is a saved set of instructions that teaches your AI assistant one research task, such as comparing a company with its peers. The pack includes company snapshot, peer comparison and segment breakdown, plus a get-started skill for your first company lookup and a setup guide for connecting. Each answer shows the filing and period behind every figure.

[Setup steps](https://kabuken.intellics.ai/docs/en/api/connecting/?utm_source=intellics_news&utm_medium=news&utm_campaign=skill_pack&utm_content=news-item)

Source: EDINET (FSA Japan), PDL 1.0; data processed by Kabuken.

Kabuken skills for Claude: research routines you install once

Claude comparing three Japanese companies side by side, each figure with its period and source filing.

kabuken-skill-pack-claude-peer-comparison-table-wide.webp

kabuken-skill-pack-claude-peer-comparison.webp

Kabuken, our platform for the financials of Japanese listed companies, is now listed in Anthropic's Claude connector directory.

Add it to Claude, then ask about a Japanese company in plain language. The figures come from EDINET filings, the disclosure system of Japan's Financial Services Agency. For each figure, Kabuken returns the source filing. Labels are in English and Japanese.

You can look up financial statements, business and geographic segments, major shareholders and the filing list for a company. You can screen by ROE, ROA, operating margin, current ratio and debt-to-equity, and read industry overviews.

Example, fiscal year ended 31 March 2026: Toyota revenue of ¥50.68 trillion, operating income of ¥3.77 trillion, and a dividend of ¥95 per share (¥90 the year before).

The connector is read-only. It does not provide stock prices or investment advice.

It works in Claude on the web and mobile, Claude Desktop, Claude Code and the Claude API, through the Kabuken MCP server. Sign in with Google or email.

[See the listing in the Claude directory](https://claude.ai/directory/kabuken?utm_source=intellics_news&utm_medium=referral&utm_campaign=listing_anthropic) | [Setup guide](https://kabuken.intellics.ai/docs/en/api/connecting/?utm_source=intellics_news&utm_medium=referral&utm_campaign=listing_anthropic)

Source: EDINET (https://disclosure2.edinet-fsa.go.jp/), PDL1.0 (https://www.digital.go.jp/resources/open_data/public_data_license_v1.0). Processed by Kabuken.

Kabuken is now in Anthropic's Claude connector directory

Diagram: a question goes to Kabuken, which returns a figure with its source filing

kabuken-claude-directory-tall.svg

kabuken-claude-directory.svg

We have published a case study on Kabuken, our platform for fundamental analysis of Japanese equities, built on EDINET XBRL filings.

It covers two engineering decisions in depth — auditing database indexes against real query patterns, and treating filings as history rather than a cache — and the spend ceiling that guards our own production jobs.

[Read the case study](/case-studies/kabuken)

Intellics builds AI services and mission-critical software, and helps your systems run reliably, quickly and securely.

News and announcements

New article: compare a sector from its filings with Claude and Kabuken MCP

Kabuken is now listed on mcpservers.org

Kabuken now draws a filing's income flow inside Claude

Kabuken is now in Anthropic's Claude connector directory

Kabuken skills for Claude: research routines you install once

Case study: how we built Kabuken

Partners

MLT.ai