#Japan #Generative AI #Tokyo #Moji #AI Evaluation Cloud

Optimizing AI Utilization with Real-Time Insights

In today’s rapidly evolving digital landscape, the implementation of generative AI is gaining momentum across various industries. While tools like Copilot and Gemini have become commonplace, many organizations find themselves stalled at the initial phase of integration. This stagnation often leads to several challenges, including a lack of comprehensive understanding of usage patterns, insufficient handling of sensitive information in prompts, and untapped potential for cost optimization due to issues like hallucination in generative outputs.

To address these pressing issues, Moji Inc. has introduced a cutting-edge solution designed to provide seamless integration and real-time analysis of generative AI tools. This initiative revolves around the AI Evaluation Cloud, a service that unifies log collection, KPI analysis, and risk detection to allow businesses to manage their AI utilization with precision.

Background and Challenges of Generative AI Integration

Despite the excitement surrounding the adoption of generative AI, many implementations exhibit a tendency to linger at the basic stage. Organizations struggle to quantify the real-time efficacy of these tools, often relying on qualitative insights that can be misleading. Moreover, concerns over security arise when employees inadvertently input confidential information into AI prompts, creating an urgent need for precautionary measures. Additionally, the growing phenomenon of hallucination, where AI produces unreliable outputs, adds another layer of complexity, hindering effective cost management strategies.

In projects Moji has supported, the necessity for a Software as a Service (SaaS) platform that automates logging and evaluation has become apparent. The mission is clear: enable organizations to comprehensively understand their AI tool usage while simultaneously mitigating risks.

Features of Moji’s AI Evaluation Cloud

The AI Evaluation Cloud boasts several essential features aimed at enhancing AI management:

1. Comprehensive Log Collection and KPI Analysis

• - Moji’s system automatically links with existing tools such as Copilot and Gemini, facilitating seamless data collection through SDK and API connectivity.
• - Visualization of departmental usage rates, satisfaction levels, and token consumption across applications offers valuable insights.

2. Real-time Risk Detection

• - The platform includes robust features for data loss prevention and output verification, effectively monitoring the input of confidential data.
• - Automatic flagging of inappropriate outputs or potential compliance violations enhances organizational security.

3. Utilization and Precision Assessment

• - Custom metrics reveal critical statistics like hallucination rates, response adoption rates, and response times, guiding organizations in pinpointing barriers to effective utilization.
• - Recommendations for departmental improvements based on these metrics empower companies to enhance their AI usage.

4. Cost Optimization Support

• - A comparison of target accuracy against cost metrics helps identify optimal models and plans, leading to substantial savings.
• - Moji has already achieved a maximum cost reduction of 30% through simulations aimed at reducing redundant token usage.

Subscription Pricing Plans

Moji offers several pricing tiers to cater to a variety of organizational needs:

• - FREE Plan: Offered at no cost for a departmental one-month trial, including requirement analysis and comprehensive implementation quotation.
• - Minimum Plan: Starting from ¥300,000, this plan employs a usage-based billing structure for log collection and includes simplified security assessments and cost-cutting proposals.
• - Full Implementation Plan: Beginning at ¥1,000,000, it covers customizable features, proprietary assessment criteria, and dedicated support staff for effective operations.

Future Prospects

Looking ahead, Moji plans to integrate external Business Intelligence (BI) tool APIs by the end of 2025, fully connecting existing dashboards. Additionally, Moji aims to expand support for the five major LLM vendors—OpenAI, Anthropic, Google, Meta, and Microsoft—systematically, and plans to have 500 organizations onboarded by the end of the fiscal year 2026, especially in sectors such as finance, manufacturing, and healthcare.

Executive Commentary

According to Moji’s CEO, Riki Kamano, “The introduction of generative AI is not the end goal, but merely the starting point. We are committed to preventing the 'set-and-forget' mindset surrounding AI usage by offering a dashboard that enables data-driven operational improvement and secure management. I am confident our AI Evaluation Cloud will propel organizations to the next phase of AI application.”

Company Overview

Company Name: Moji Inc.
Location: 2nd Floor, Kuwano Building, 6-23-4 Jingumae, Shibuya, Tokyo
CEO: Riki Kamano
Business: Development and consulting of generative AI solutions; AI evaluation SaaS services
URL: Moji Inc.

Contact Information

For inquiries regarding the AI Evaluation Cloud, please reach out to:

• - Email: [email protected]

The names of companies and products mentioned herein are trademarks or registered trademarks of their respective owners. Prices indicated do not include taxes.