AlpacaTech to Provide 'MixSeek AI Data Lake' Integrating Anthropic's Financial AI Agent Skills and Proprietary Platform
AlpacaTech (SBI Group) announced the construction of 'MixSeek AI Data Lake,' which adapts Anthropic's financial AI agent technology for the Japanese stock market using TDnet and EDINET data. It enables financial institutions to perform advanced PoCs for tasks like financial analysis and competitor comparison using public information without depending on external data vendors.
📋 Article Processing Timeline
- 📰 Published: May 18, 2026 at 22:00
- 🔍 Collected: May 18, 2026 at 13:31
- 🤖 AI Analyzed: May 18, 2026 at 13:45 (13 min after Collected)
AlpacaTech Inc. (Headquarters: Chiyoda-ku, Tokyo; CEO: Seibun Yotsumoto), a subsidiary of FOLIO Holdings which provides innovative financial solutions within the SBI Group, constructed the 'MixSeek AI Data Lake' on May 15. This AI-ready data platform ports various skills from Anthropic's financial AI agent reference implementation, 'financial-services,' for use with Japanese stock data, including disclosures from TDnet and EDINET.
On May 5, 2026, Anthropic released AI agents for the financial industry. AlpacaTech rapidly prepared a platform applicable to Japanese stock data using its proprietary 'MixSeek' infrastructure. This allows financial institutions and corporations to conduct advanced PoCs of AI agents utilizing skills released by Anthropic.
The platform features a highly optimized, AI-ready data infrastructure that enables AI agents to refer to public information—such as quarterly earnings reports, securities reports, and corporate governance reports from TDnet and EDINET—in a fast and scalable manner. Notably, it allows for PoCs based primarily on public information without depending on contracts with external financial data vendors, as assumed by Anthropic's skills.
## Development Background
In the financial industry, expectations are rising for the use of generative AI and AI agents in tasks such as corporate research, earnings analysis, report creation, and investment hypothesis testing. Anthropic's 'financial-services' provides design examples for agents and skills supporting multiple financial workflows, including investment banking, research, and wealth management.
For Japanese stock analysis, PDF disclosures from TDnet and EDINET are crucial primary sources. Stable reference and analysis of these documents require unique infrastructure for PDF parsing, text extraction, metadata attachment, and indexing. AlpacaTech’s 'MixSeek' is a technical platform that combines multiple models like ChatGPT, Gemini, Claude, and Grok to achieve sophisticated analysis results.
## Available AI Agent Skills for PoC
The platform includes skills ported for Japanese stock and language disclosures based on Anthropic's workflows:
- **Financial Analysis**: Retrieves quarterly reports, organizes forecasts vs. results, and segment performance in Japanese.
- **Morning Note Creation**: Searches cross-sectionally for TOPIX 500 company disclosures on a specific day and summarizes noteworthy items.
- **Cross-Sector Analysis**: Organizes performance, risks, and strategies for companies within the same sector based on industry classifications.
- **Investment Hypothesis Verification**: Updates the continuity of hypotheses and risk factors based on new disclosures.
- **Competitor Analysis**: Identifies strategies and risks from securities reports of peers and performs comparisons.
On May 5, 2026, Anthropic released AI agents for the financial industry. AlpacaTech rapidly prepared a platform applicable to Japanese stock data using its proprietary 'MixSeek' infrastructure. This allows financial institutions and corporations to conduct advanced PoCs of AI agents utilizing skills released by Anthropic.
The platform features a highly optimized, AI-ready data infrastructure that enables AI agents to refer to public information—such as quarterly earnings reports, securities reports, and corporate governance reports from TDnet and EDINET—in a fast and scalable manner. Notably, it allows for PoCs based primarily on public information without depending on contracts with external financial data vendors, as assumed by Anthropic's skills.
## Development Background
In the financial industry, expectations are rising for the use of generative AI and AI agents in tasks such as corporate research, earnings analysis, report creation, and investment hypothesis testing. Anthropic's 'financial-services' provides design examples for agents and skills supporting multiple financial workflows, including investment banking, research, and wealth management.
For Japanese stock analysis, PDF disclosures from TDnet and EDINET are crucial primary sources. Stable reference and analysis of these documents require unique infrastructure for PDF parsing, text extraction, metadata attachment, and indexing. AlpacaTech’s 'MixSeek' is a technical platform that combines multiple models like ChatGPT, Gemini, Claude, and Grok to achieve sophisticated analysis results.
## Available AI Agent Skills for PoC
The platform includes skills ported for Japanese stock and language disclosures based on Anthropic's workflows:
- **Financial Analysis**: Retrieves quarterly reports, organizes forecasts vs. results, and segment performance in Japanese.
- **Morning Note Creation**: Searches cross-sectionally for TOPIX 500 company disclosures on a specific day and summarizes noteworthy items.
- **Cross-Sector Analysis**: Organizes performance, risks, and strategies for companies within the same sector based on industry classifications.
- **Investment Hypothesis Verification**: Updates the continuity of hypotheses and risk factors based on new disclosures.
- **Competitor Analysis**: Identifies strategies and risks from securities reports of peers and performs comparisons.
FAQ
What can I do with AlpacaTech's AI Data Lake?
You can use AI to automatically parse TDnet and EDINET disclosures for instant financial summaries, competitor comparisons, and investment hypothesis testing.
Why is Anthropic's technology used?
Anthropic has released advanced AI agent skills for the financial sector. Porting these to Japan allows for world-class analytical capabilities in the local market.
Is a contract with an external data vendor required?
No. Since the platform primarily uses public information, you can start PoCs without individual contracts with external data vendors.