Launch of a Continuous Improvement Model to Support the Transition of Conventional SaaS to Production AI and AI Agents

Headwaters Co., Ltd. (Headquarters: Shinjuku-ku, Tokyo; CEO: Yosuke Shinoda), a company in the AI solutions business, and Headwaters Consulting Co., Ltd. (Headquarters: Shinjuku-ku, Tokyo; CEO: Kenji Kato) have begun offering a continuous improvement cycle model. This model utilizes AI implementation and improvement assets accumulated from multiple projects to support the shift of conventional SaaS to production AI and its transformation into AI agents (*1).

This model is designed to first embed AI functions, centered around RAG (Retrieval-Augmented Generation) (*2), into SaaS customer touchpoints such as inquiry handling, knowledge search, and proposal support with production-level quality, facilitating an early market launch. It then provides a framework for gradual advancement by repeatedly making improvements and expansions based on usage status, customer reactions, and profitability.

• Background

In recent years, corporate AI utilization has expanded beyond improving internal operational efficiency to enhancing the value of SaaS and digital services that customers use directly. There is a growing need for new service experiences using AI agents and RAG, especially at customer touchpoints such as inquiry handling, product/service recommendations, knowledge searches, and business process support, which have traditionally relied on human or screen-based operations.

On the other hand, while SaaS operating companies want to improve the quality of answers and convenience for users by incorporating AI into their existing services, an increasing number of them face challenges in improving accuracy, preparing data, controlling authentication/permissions, and establishing an operational structure required for production deployment. Particularly in production services used by customers, unlike the PoC stage, quality is required that includes not only response accuracy but also safety, reproducibility, and a monitoring system.

In the medium to long term, it is anticipated that the use of AI agents will evolve SaaS from merely providing a screen to becoming a "task-performing service" that handles everything from search and response to business process support and integration with external systems. However, while the use of AI agents is powerful, the difficulty of implementation and operation increases in stages. Therefore, it is crucial to first build a RAG/AI foundation that can withstand production use and then have a structure that is easily expandable to AI agent utilization.

Given this background, companies are seeking a continuous improvement support model that not only adds AI functionality but also safely and accurately enhances the SaaS or customer-facing services themselves, allowing for a phased evolution from RAG to AI agent utilization.

In response to these market changes, Headwaters and Headwaters Consulting will first implement and operate high-quality AI functions centered on RAG for production services used by customers. By providing a structure that can easily be connected to subsequent AI agent utilization, we will support companies in enhancing their customer experience and creating new business value. In addition to implementation support, we will also assist in building a system where client companies can eventually run the improvement cycle internally through knowledge transfer and hands-on support.

• Features of the Support Provided

As a technology foundation, we will utilize Microsoft Foundry (*3) and provide integrated support covering everything from building a knowledge base combining Foundry IQ and AI Search, to the high-speed implementation of agent-type RAG, continuous accuracy improvement (RAGOps) (*4), evaluation/observability design for AI functions, and governance/monitoring operations.

① Mi...