ELYZA
: How to use it, features, and the business problems it solves
Add bookmark
What is ELYZA?
ELYZA provides generative AI (LLM) adoption support for large enterprises from ELYZA, Inc., part of the KDDI Group. It covers identifying business use cases, building processing flows and APIs, fine-tuning, and developing company- and industry-specific LLMs. Pricing is not disclosed (individual quote). Cases include cutting after-call work time by about 50% at JR West Customer Relations (announced March 2026).
Business problems it solves
About "ELYZA"
What is ELYZA?
ELYZA is an AI company that researches and develops its own large language models (LLMs) and uses that expertise to support large enterprises in adopting generative AI and developing industry-specific models. It is run by ELYZA, Inc., founded in September 2018 by founders from the Matsuo Laboratory at the University of Tokyo, and belongs to the KDDI Group.
Generative AI adoption support is offered to large enterprises as the "AI Research & Solution business." It ranges from practical deployment support that embeds general-purpose models such as GPT into business operations, to fine-tuning (specialized training), to developing LLMs specific to a company or industry. Its foundation is the technical capability behind the Japanese "ELYZA LLM" series it has developed and released, and it has many cases in text-heavy work such as contact centers and in credit review work in finance.
Engagement overview
How the engagement proceeds
ELYZA does not publish a standard stage-by-stage flow. Its services page describes practical deployment support as "consistent support from identifying the business scenes where generative AI should be used, to building processing flows that raise output quality and developing APIs." Published cases suggest the following progression.
-
Identify business scenes
Determine which work generative AI should be used for. With JR West Customer Relations, a joint project began in 2022 and targeted the summarization of inquiries.
-
Build processing flows, develop APIs, specialize models
Design processing flows that raise output quality and, where needed, fine-tune or develop specialized models. Security and generation speed requirements are also met at this stage.
-
Deploy on site
Embed the system into real operations. The official case introduction describes the JR West Customer Relations summarization AI as taking "only two months to deploy."
-
Operational setup and continuous improvement
Both companies keep improving functions and operations after rollout. Following improvements since its 2023 introduction, the client cut after-call work time by about 50% in an operation of around 100 people (announced March 2026).
Service menu
01Practical deployment of generative AI
Support for embedding general-purpose models such as GPT into business operations.
- Identifying business scenes — Determining which work generative AI should be used for
- Processing flows and API development — Building mechanisms that raise output quality and integrating with systems
02Specialized training and model development
Support for requirements that general-purpose models alone cannot meet.
- Specialized training support for generative AI — Fine-tuning proprietary or open models to meet security, generation speed and industry-specific requirements
- Company- and industry-specific LLM development — Developing the model itself to handle industry-specific tasks and terminology
03Products (AI Product business)
In-house products used together with custom development.
- ELYZA Works — A business tool that lets frontline teams create and share company-specific AI apps just by describing the app they want. Resold by KDDI as "ELYZA Works with KDDI"
- ELYZA Voice Agent — A voice dialogue AI agent that carries tasks such as reservations, changes and identity verification through to completion according to business rules (announced October 2026, being validated for deployment)
In an official note post in March 2026, ELYZA said its solution business would minimize one-off custom development and provide applications within the scope of "ELYZA Platform," a base with shared functions. In domains that require deep industry knowledge, such as finance, manufacturing, healthcare and government, it plans to provide vertically integrated support including specialized models.
Pricing
Prices are not published. They are not disclosed (individual quote) as of October 2026. No reference amounts for PoCs, joint research or specialized model development are listed on the official site.
Track record and case studies
01Contact centers and customer service
Deployment for summarizing calls and inquiries and drafting responses.
- JR West Customer Relations — Made call summarization up to 54% more efficient. Later operational setup and model improvements cut after-call work time by about 50% in an operation of around 100 people (announced March 2026)
- Tokio Marine & Nichido — Cut the effort of drafting customer response text by 50%
- Altius Link and KDDI — Jointly developed a contact-center-specific LLM app and a contact reason classification LLM app
02Finance and review
Automating work that involves judgment.
- Sumitomo Mitsui Card — Began using an automated decision AI for membership screening, automating 20% of reviews previously done by staff (announced July 2026)
- Meiji Yasuda Life — Introduced ELYZA's generative AI service at its communication center (announced January 2024)
03Text generation and government
Use in drafting and in public administration.
- Mynavi — Made job posting drafting 30% more efficient. Provided technology for and co-developed "Mynavi AI Pencil"
- Digital Agency — Commissioned for "technical validation and environment setup for the business use of generative AI" (disclosed June 2025). Selected with KDDI as a "domestic LLM for trial use in Government AI" (announced March 2026)
The official site also displays logos of SmartNews, Sompo Japan, Deloitte Tohmatsu, Mitsubishi UFJ Research and Consulting, Mori Hamada & Matsumoto and others.
Who it suits and who it may not
Good fit
- Large enterprises that want to embed AI into text-heavy work such as contact centers, or into review work
- Companies that need more than a general-purpose model API because of confidentiality or speed requirements and are considering specialized models
- Companies that must handle industry-specific terminology and knowledge, such as finance, healthcare and government
May not fit
- Small and medium-sized businesses seeking custom solution support (the company states it is for large enterprises)
- Companies whose main goal is employee training or company-wide AI talent development (no training menu is listed officially)
- Companies that want a price estimate before inquiring
FAQ
Q. How much does ELYZA's generative AI adoption support cost?
Prices are not published; quotes are individual. No reference amounts for PoCs or specialized model development are listed on the official site (as of October 2026).
Q. Can small and medium-sized businesses use it?
The solution business is officially described as "for large enterprises." The business tool "ELYZA Works," which lets frontline teams build AI apps, is offered as a separate product.
Q. Can ELYZA build an LLM for our company?
The official services include "company- and industry-specific LLM development" and "specialized training support for generative AI (fine-tuning)." In healthcare, it has developed the Japanese medical LLM "ELYZA-LLM-Med."
Q. What is the relationship between ELYZA and KDDI?
A capital and business alliance was announced on March 18, 2024. The announcement stated that, around April 1, 2024, KDDI would hold 43.4% and KDDI Digital Divergence Holdings 10.0% of ELYZA's shares, making ELYZA a consolidated subsidiary of KDDI. ELYZA's official site now displays the KDDI Group's policies.
The information on this page is based on what was published on the official website as of October 2026. Please check the official website for the latest pricing and details.
