Announcing the launch of the on-premise AI solution "SapiaBox

Ascentec Corporation (headquartered in Chiyoda-ku, Tokyo; President: Takashi Matsuura; hereinafter "the Company"), an AI application development and on-premise AI infrastructure provider, has partnered with "closip" (headquartered in Minato-ku, Tokyo; President: Takehiro Ikeda; hereinafter "closip") to launch an on-premise AI solution, SapiaBox, on April 22, 2026. (headquartered in Minato-ku, Tokyo; Takehiro Ikeda, President; hereinafter "closip"), a provider of on-premise AI infrastructure, and started offering the AI solution "SapiaBox" in on-premise environment on April 22, 2026.

Background

In recent years, the use of generative AI has been rapidly expanding in all industries. On the other hand, the following three major issues have been pointed out in the operation of AI on cloud computing.

Security risk in handling personal and confidential information

Response delays due to the Internet

Increased operational costs due to pay-as-you-go billing

These issues have become barriers to the use of AI in organizations that handle highly confidential information, such as medical institutions, government agencies, financial institutions, and professional firms.

What is SapiaBox?

SapiaBox is an on-premise AI solution developed by closip to solve the above-mentioned problems of generated AI. SapiaBox is not a product that depends on specific hardware, but rather an end-to-end solution service that provides a full range of services from the selection of the most appropriate LLM (Large Language Model), development of business applications, and implementation in an on-premise environment after listening to the customer's business issues.

The closed environment configuration, which does not transmit any data to the cloud, minimizes the risk of information leaks, while realizing a cost structure that does not incur additional charges even if the number of users increases.

SapiaBox Introduction Flow

SapiaBox is deployed in the customer's environment according to the following flow.

SapiaBox implementation flow

1. hearing (requirement definition)

We understand the customer's business issues and select the optimal hardware and LLM model, including our Edge AI Array. 2.

2. application development and environment construction

Based on the results of the hearing, we develop an AI application optimized for the customer's business and build an on-premise environment. 3.

3. installation and configuration

We configure the connection and security settings according to the customer's network environment, and deliver the application ready for use. 4.

Training (AI application in-house production training) *Optional

We provide practical training to enable customers to develop and improve AI applications by themselves (see below for details). 5.

5. start of operation

Full-scale operation is started.

Of the above, CLOSIP will provide software services (1-4), and MHI will be in charge of hardware procurement, maintenance, and support.

Significance of the alliance

We have invested in the company and jointly provided the secure remote access service "Resalio Connect" with them.

This partnership in the field of AI solutions builds on this existing relationship of trust.

The quality of a generative AI solution is assured only when both software (AI model selection and application development) and hardware (on-premise infrastructure that operates stably) are in place. This partnership is possible because we have our own on-premise AI platform "Edge AI Array" and have been providing security solutions for over 20 years.

SapiaBox Hardware Lineup

SapiaBox allows customers to select the most suitable hardware depending on their scale of use as shown below, both of which are provided by our one-stop service from procurement to maintenance.

For small to medium scale and PoC use: "NVIDIA DGX Spark" and equivalent products

Compact and ideal for departmental AI applications and prototype verification.

NVIDIA DGX Spark and equivalent products

For more information on SapiaBox

https://www.ascentech.co.jp/solution/sapiabox/sapiabox.html

For large scale and multiple users: Our original product "Edge AI Array

Up to five NVIDIA GPUs for AI are installed in a 2U chassis. It is an on-premise AI platform for large-scale LLM, virtual humans, and various AI applications.

Edge AI Array

For more information on the Edge AI Array, please visit

https://www.ascentech.co.jp/solution/eaa/eaa.html

SapiaBox offers the following value propositions

1. optimal LLM selection and custom application provision

SapiaBox selects the most appropriate LLM model according to the customer's business issues, and develops and implements a custom application that combines elemental technologies such as RAG (search extension generation) and OCR. By providing an "AI application that fits the customer's business," rather than introducing a generic chat tool, we create an environment that is effective in practice immediately after its introduction.

2. support for in-house production of AI application (development training)

As an option, SapiaBox provides a practical training program to enable customers to develop and improve their own AI applications.

In-house production of AI applications will enable you to reflect issues in the field in your applications in real time, which will lead to a higher retention rate of AI applications and a significant return on investment. AI applications can be created in-house to reflect issues in the field in real time, significantly improving the retention rate and return on investment of AI utilization.

The training will be conducted in a practical format using the actual SapiaBox environment, allowing participants to systematically learn everything from the basics to application.

Takehiro Ikeda, President of closip, comments

Many companies and organizations are facing this very challenge: "We want to introduce generative AI, but we don't know where to start.

SapiaBox is more than just an AI server. SapiaBox is more than just an AI server; it is a solution that understands your business, designs and develops optimal AI applications, and accompanies you until you are ready to use AI yourself.

We are very pleased to have partnered with Asentech, which has its own on-premise AI infrastructure and a long track record in the security field. With both software and hardware in place to support our customers, we will provide an environment in which AI can be used with confidence in fields that require high information security, such as healthcare, government, and finance.

About Ascentec Corporation

Ascetec is a company that provides a secure environment for the use of AI in fields requiring high information security, such as medicine, government, and finance! Contributing to the transformation of our customers' business workstyle. Under the corporate philosophy of "Contributing to our customers' business workstyle transformation," the company's main businesses are the development, sales, and consulting services of virtual desktop-related products, as well as cyber security solutions and cloud service-related businesses. In addition, we will strengthen our ESG initiatives and contribute toward a sustainable society.

Listed on the Tokyo Stock Exchange Standard Market [Securities Code: 3565].

*Company names and product names mentioned herein are trademarks or registered trademarks of their respective companies.