LiberCraft Co., Ltd. Provides Generative AI and RAG Consulting for Major Tokyo Broadcaster

LiberCraft provided hands-on technical consulting for a major Tokyo broadcaster, helping improve their internal AI chatbot's accuracy and establish an evaluation framework to foster in-house AI capabilities.
提携NQ 83/100出典:PR Times

📋 Article Processing Timeline

  • 📰 Published: May 22, 2026 at 17:30
  • 🔍 Collected: May 22, 2026 at 09:01
  • 🤖 AI Analyzed: May 22, 2026 at 09:03 (1 min after Collected)
## LiberCraft Co., Ltd. Provides Generative AI and RAG Consulting for Major Tokyo Broadcaster

LiberCraft Co., Ltd. (Headquarters: Tokyo; CEO: Daigo Miyoshi) announced that it provided consulting support for generative AI and RAG technologies to a major Tokyo broadcaster from January to March 2026. The project, conducted over nine weekly sessions, contributed to improving the precision of their internal AI chatbot and establishing a foundation for continuous improvement.

### Background and Significance

The major Tokyo broadcaster had already implemented an AI chatbot to improve operational efficiency. However, they faced challenges with response precision and needed a methodology for improvement and a system for evaluation.

LiberCraft addressed these challenges by providing hands-on consulting leveraging their expertise in RAG (Retrieval-Augmented Generation) and LLMs.

### Overview of Support

This project was not an outsourced system development contract but a weekly technical consulting project. While the broadcaster's technical team led the initiative, LiberCraft provided professional expertise and support, creating a structure that promoted in-house capabilities while delivering practical results.

- **Formulating a RAG Pipeline Evaluation Method**: Constructed a framework to systematically analyze challenges in response precision and provide quantitative evaluation.
- **Support for Search and Response Precision Improvement**: Assisted in improving search precision and response quality through data structuring and optimization of prompt design.
- **Advanced Proposals for the Next Phase**: Presented architectural proposals and a roadmap for further precision improvements.

### LiberCraft's Features and Strengths

- **Proven Track Record in Advanced and Complex Generative AI Development**: Beyond simple tool implementation, LiberCraft has accumulated extensive experience supporting complex RAG and local LLM development.
- **Capability Across Broad AI and Data Science Technologies**: From classical data analysis and predictive modeling to state-of-the-art generative AI development, they provide optimal methods based on client goals.
- **Hands-on Support for In-house AI/Data Utilization Through Training**: They support not only AI system development but also the training of personnel to utilize AI and data, ensuring true in-house capability and continuous utilization.

### Comment from Representative Director Daigo Miyoshi

"In internal generative AI utilization, the biggest barrier when moving from PoC to production is establishing a methodology to quantitatively evaluate and continuously improve precision."

"By discussing weekly with the team at the major Tokyo broadcaster and providing end-to-end support from evaluation framework construction to concrete improvement approaches, this effort embodies our vision of being a 'technical partner, not just an outsourcing firm.'"

"We will continue to accelerate the social implementation of generative AI through technical consulting that supports our clients' autonomous improvement cycles and internal capacity building."

FAQ

株式会社リベルクラフトはどのようなコンサルティングを提供しましたか?

生成AI・RAG技術を活用し、社内AIチャットボットの回答精度向上、評価フレームワークの構築、継続的な改善基盤の構築を支援する伴走型技術コンサルティングを提供しました。

このプロジェクトはどのような座組みで行われましたか?

システム開発の受託ではなく、週次定例の技術コンサルティングとして実施されました。放送局側の技術チームが主体的にプロジェクトを推進する中で、リベルクラフトが専門的知見の提供とサポートを行う形です。

回答精度の課題に対してどのようなアプローチがとられましたか?

体系的な分析に基づき定量的に評価できるフレームワークを構築し、データの構造化やプロンプト設計の最適化を通じて検索・回答精度の改善を図りました。

リベルクラフトのAI開発支援における強みは何ですか?

高精度なRAG開発やローカルLLM導入などの複雑な技術実績、古典的データ解析から生成AIまでの幅広い対応力、そして人材育成による内製化支援を強みとしています。

代表の三好大悟氏は、AIの実運用における重要ポイントをどう見ていますか?

PoCから実運用に進む際の最大の壁は「精度をいかに定量的に評価し、継続的に改善していくか」という方法論の確立にあると指摘しています。