DATAFLUCT Announces Autonomous AI Agent "Airlake BI Agent v2" for Natural Language Search, Analysis, and Document Creation from Dispersed Customer DBs and Proposal Materials within Companies

DATAFLUCT Inc. announces "Airlake BI Agent v2," an AI analysis agent that autonomously completes customer analysis and proposal material preparation by integrating and analyzing various dispersed data within companies, such as customer information and proposal documents, using natural language.

This service is a significant update based on the knowledge gained from the beta version of "Airlake BI Agent" released in 2025. It liberates sales representatives from administrative tasks such as market information gathering, complex sales/customer data analysis, and proposal document creation, allowing them to shift their focus to the essential task of value proposition.

Background: Dispersed Information and the Burden of Peripheral Tasks in Sales

In recent years, while many companies have been implementing CRM/SFA systems, sales sites still spend a considerable amount of time on peripheral tasks such as information gathering, document creation, and data analysis.

For example,

Searching for information from vast product databases

Searching for and reusing past proposal documents

Analyzing sales performance and inventory data

These tasks, while essential for sales activities, require significant time and effort, encroaching on the customer communication and proposal activities that sales representatives should be focusing on.

Furthermore, the dispersion of necessary information across multiple systems and files, including CRM/SFA, core systems, proposal documents, PDFs, and paper documents, has led to issues such as "inability to quickly access necessary information," "data analysis being siloed to specific individuals," and "past knowledge not being utilized."

To address these challenges, DATAFLUCT has developed an AI agent that comprehensively integrates and analyzes CRM/SFA, core data, and internal knowledge to support sales activities and decision-making.

Dispersed Customer/Internal Data Transformed into Actionable Insights for Sales

Airlake BI Agent Screen Image

"Airlake BI Agent v2" supports sales activities and management decisions with four functions: "Search," "Analyze," "Create," and "Manage."

1. Information Search: Search dispersed customer information using natural language

Search for necessary information using natural language from SFA/CRM, business DBs, core systems, POS, Excel, meeting minutes, proposals, product catalogs, and more.

For example, sales representatives and sales planning staff can start gathering information needed for customer-specific reviews by asking questions like, "Summarize the recent negotiations, sales performance, and past proposals for Company A."

2. Analysis: Organize What / Why / Next Steps

Organize the factors behind changes and candidates for next actions based on sales, gross profit, reasons for lost deals, proposal history, customer-specific sales data, etc.

The system is designed to clarify "what to do next" for the numbers visible in existing BI tools by connecting structured data with related documents and allowing analysis while confirming the basis.

3. Document Creation Support: Translate analysis results into proposal preparation

Connect analysis results to business outputs such as sales materials, proposal preparation memos, customer-specific reports, and next action proposals, rather than just ending with analysis.

For instance, when requested to "Prepare for proposing product XX to Company A," it supports summarizing the proposal potential, anticipated discussion points, and past similar proposal documents.

4. Knowledge Management: Utilize past proposals, FAQs, and meeting minutes

Treat past proposals, FAQs, meeting minutes, sales memos, success stories, etc., as knowledge that can be easily searched and reused.

This allows for the organizational utilization of sales expertise and proposal patterns that tend to become siloed, enabling even less experienced employees to engage in sales activities that leverage past knowledge immediately.

【Use Cases】

Managers: Transition from a situation requiring several days for cross-sectional analysis like "Trends in sales and gross profit margins by customer and product" to immediate understanding of figures and consideration of next actions.

Sales Planning: Significantly reduce the time and effort required for analyzing market/own product trends and customer-specific sales trends, and for considering sales strategies. By simply asking the AI in natural language, sales strategies can be formulated in line with the latest situation.

Sales Representatives: Eliminate the hassle of searching for optimal slides from vast past proposal documents. For example, the AI can extract "configurations that had a high closing rate for the manufacturing industry," thereby improving the quality of proposals.

Inquiries about "Airlake BI Agent v2"

Airlake BI Agent is an AI analysis agent that supports data extraction, visualization, cause analysis, and organization of next actions using natural language across Excel, SFA/CRM, POS, core systems, meeting minutes, and more. Analysis can begin with existing data without writing SQL or Python.

For service details and implementation consultations: https://datafluct.com/ai-service/ai-agents/bi-agent/

Examples of DATAFLUCT's AI Agent Implementation

DATAFLUCT's AI agents have been introduced in various forms to major domestic companies, including the implementation of a "Sales Support AI Agent" at Yamasei Co., Ltd., a major confectionery wholesaler. The project with Yamasei involved integrating product information, vast sales data, promotional materials, and sales knowledge accumulated in the field. Verification is underway to utilize conversational search, extraction of key points from documents, and AI-driven company analysis in actual sales activities.

(Press release announced March 2026 https://prtimes.jp/main/html/rd/p/000000337.000046062.html)

The current "Airlake BI Agent v2" standardizes and expands the "AI usage truly demanded by the field," as confirmed in these pioneering cases, into product features.

Proprietary Technology "CIE" Connecting Numbers and Context, and Robust Governance Compliant with Large Enterprises

"Airlake BI Agent v2" adopts an advanced multi-agent configuration where multiple specialized AI agents autonomously collaborate, from intent extraction to data exploration, analysis, quality checks, and output integration.

By combining the proprietary technology "CIE (Contextual Inference Engine)" with schema grounding, a semantic layer is constructed, enabling advanced fusion of numerical data (What) and text information such as meeting minutes (Why). This thoroughly suppresses AI-specific incorrect answers (hallucinations) and allows for analysis based on solid evidence.

Furthermore, it complies with the strict IT governance standards of large enterprises through read-only (reference only) connections, detailed permission designs, and retention of operation logs. It safely and effectively utilizes the company's internal information assets as they are, without replacing existing IT environments such as CRM/SFA and core ERP.

Future Outlook: Democratization of Data Utilization, Towards an Era of "One Person, One AI Analyst"

In the future, DATAFLUCT will promote the development of the next version, "v3," which will provide dedicated AI agents optimized for individuals, in addition to shared environments for companies and departments. With a one-user-one-agent configuration, the goal is to realize "one person, one AI analyst" that supports the business context of each individual sales representative and empowers decision-making daily through the power of data.

Comment from Hayato Kumemura, Representative Director and CEO of DATAFLUCT Inc.

Corporate data utilization is moving from the stage of viewing numbers on dashboards to the stage of confirming "why" and "what to do next" with the same evidence in meetings. On the other hand, there is still much work remaining to convert the information accumulated in SFA/CRM and business databases into terms that the field can understand. Airlake BI Agent v2 supports the creation of an environment where the field and management can organize customer-specific discussion points while leveraging the existing data environment. DATAFLUCT will continue to refine the infrastructure and business implementation for transforming data into business, together with our customers.

About DATAFLUCT Inc.

DATAFLUCT Inc., with the vision of "Data into Business," is a data business partner that creates new value from buried data and solves social issues. With strengths in "multimodal data utilization" that is not constrained by data format, including unstructured data, we achieve data collection, accumulation, processing, and analysis in an end-to-end manner.

We also offer proprietary services that focus on global-standard challenges such as reducing waste through demand forecasting, sustainable urban planning, and behavioral change towards decarbonization, aiming to realize a world where everyone can make sustainable decisions by effectively utilizing data. Certified as a JAXA Venture* in 2019.

*A venture company established and funded by Japan Aerospace Exploration Agency (JAXA) employees, utilizing intellectual property and knowledge from JAXA's operations.

Head Office Location: 7F, SHIBUYA SAKURASTAGE SHIBUYA SIDE SHIBUYA TOWER, 1-4 Sakuragaoka-cho, Shibuya-ku, Tokyo

Representative: Hayato Kumemura, Representative Director and CEO

Established: January 29, 2019

Phone Number: 03-6822-5590 (Main)

Capital: 1,497,120,000 yen (including capital reserve)

Business Activities: Data platform construction and operation support, DX promotion support and operation support, Sustainable data business

Website: https://datafluct.com/

Official X: https://twitter.com/datafluct

Facebook: https://www.facebook.com/datafluct/

Inquiries regarding this matter

For service details and implementation consultations: https://datafluct.com/

Inquiries from the press: [email protected]