ABEJA Platform - Digital Version of EMS
For more than 10 years since its establishment, ABEJA has been continuously working on R&D for ABEJA Platform, as well as developing and operating AI systems for client companies. The R&D and investment in ABEJA Platform, which has been carried out since the early days of deep learning technology, has resulted in ABEJA's overwhelming technological strength and success in reaching production in the domestic market today.
ABEJA Platform is equipped with cutting-edge manufacturing machinery for the entire process of designing, developing, building, operating and improving the DX of clients' core business, and the manufacturing expertise to skilfully utilise and optimise it.
Leveraging the ABEJA Platform, ABEJA is developing a business model similar to an Electronics Manufacturing Service (EMS), seamlessly integrating AI systems into core operations and providing fully managed DX services to meet the needs of client companies.
A factory that manufactures AI systems using "state-of-the-art manufacturing machinery + manufacturing expertise".
Features of ABEJA Platform
ABEJA's unparalleled technological capabilities and achievements in reaching production are based on the "state-of-the-art manufacturing machinery" (= modules) and "manufacturing expertise" (= know-how) built within ABEJA Platform.
特徴
01
"State-of-the-art manufacturing machinery"
(= module) in ABEJA Platform
The five layers that make up ABEJA Platform
DX needs vary widely. ABEJA Platform is a software suite that provides processes from design, development, construction and operation to improvement, tailored to the needs of the client company and ensuring sustainable and stable operations. It is fully equipped with versatile, state-of-the-art modules that can be used in all processes in all directions, and is organised into 'parts' in five main layers.
Each component is pre-built and can be combined and set up to meet the needs and challenges of the client company, making it possible to provide a unique system. It would take an enormous amount of time and cost to create this from scratch if they were to build it themselves. The planning and creation of design documentation, which is the initial stage of the process, is an extremely difficult part of the DX process for companies.
It would be too difficult and impractical to simply provide the ABEJA Platform as is to client companies and let them use it on their own. Therefore, we do not currently offer traditional SaaS (Software as a Service)* or PaaS* (Platform as a Service)* where customers have to use it independently. We offer ABEJA Platform as a digital version of EMS, with ABEJA consultants and data scientists taking the lead in creating the design documents.
In addition, ABEJA Platform is highly flexible and able to build seamlessly while incorporating "human" intervention, i.e. manufacturing expertise at key stages of the process. It quickly implements efficient and robust DX processes that respond to customer needs.
*SaaS (Software as a Service): Provides software that is normally offered as a packaged product
*PaaS (Platform as a Service): Provides the complete platform required for development, including servers, hardware, operating systems, middleware and databases
BaaS※ Layer
Attribute/
motion estimation
Congestion prediction/
Face detection
Demand
prediction
Generic AI models are pre-built
Search similar image/
video
Knowledge search/
Chatbot
Anomaly
detection
Core Layer
Data generation, collection
Stores technology for creating software
Data
processing
AI
modeling
Large Language Model
Security Layer
Processing to maintain security
Encryption
Decryption
Authorization
Computing Layer
Computational processing of input data
Cloud
Edge
Connect Layer
Input the data required
Camera
Sensor
SNS
Files
*BaaS (Banking as a Service) is a cloud service that provides back-end functionality for applications.
"Manufacturing expertise” making full use of state-of-the-art manufacturing machinery (know-how)
特徴
02
“Human in the Loop (Collaboration of Human and AI)” approach that can even realize Zero PoC (Proof of Concept)
Many companies conduct meticulous PoCs when driving DX with AI. However, in many cases, projects get stuck at PoC and fail to reach production for reasons such as the AI's accuracy not improving as expected or the inability to afford consistent investment.
ABEJA advocates a Human-in-the-Loop approach that assumes human involvement. This is a mechanism for efficiently building AI models by accumulating business process expertise and knowledge as data on ABEJA platform, and having humans assist with judgement and decision-making at the necessary points.
In the initial phase, when operational expertise and data are insufficient, humans are primarily responsible for decision-making, but as the steps progress, the involvement of humans and AI can change. Customers can use AI in their business processes right from the start, and humans can nurture AI in the operational process. This accelerates the transformation to an AI-powered business process and the AI training cycle, resulting in faster and more sophisticated progression to actual live operations.
Using the Human-in-the-Loop approach to build an AI system with ABEJA Platform enables zero PoC and rapid transition to a production environment
Executed by human
STEP 1 (Before DX)
The target process
Knowledge
.png)
Operator
-
Human executes business processes in real spaces
-
Operational know-how and knowledge is distributed among individuals, people, etc.
Executed by human
STEP 2
The target process
Accumulates
data
Operator
-
Introduce ABEJA Platform for the business processes that are handled by human
-
Human handles business processes on ABEJA Platform
-
Operational know-how and knowledge are stored in ABEJA Platform as data
STEP 3
Executed by human, assisted by AI
The target process
Accumulates
data
Assisted by AI
Operator
Reduced load
-
Human executes business processes on ABEJA Platform.
-
The operational know-how and knowledge gradually accumulated in ABEJA Platform is utilised as data, AI provides assistance, and the workload of humans is reduced.
-
Data is accumulated and utilised on ABEJA Platform through daily business operations, and the assistance provided by AI becomes more sophisticated.
STEP 4
Executed by AI, assisted by human
The target process
Accumulates
data
Executed by AI
Operator
Assisted by human
-
AI executes business processes on ABEJA Platform
-
Human task is mainly supervision. The workload for human becomes even lighter.
-
Operational know-how and knowledge is used as data on ABEJA Platform, making execution even more sophisticated.
Advantages of ABEJA Platform
State-of-the-art 'manufacturing machinery' and 'manufacturing know-how' for all the processes required for DX are available here.
With ABEJA Platform, you can achieve faster implementation than with in-house development and operation. It also gives you access to advanced technology to reduce costs, avoid risks and improve capabilities.
01
Unparalleled speed from development to operation
ABEJA Platform service has already been provided to a wide variety of companies, and is equipped with “manufacturing machinery” (modules) that have been developed and proven. Modules can be selected and combined as required, significantly reducing development time compared to building the system development environment from scratch.
ABEJA's consultants, who have a wealth of expertise and in-depth knowledge, will also accompany you and propose a lean plan from the beginning, including identifying issues and areas with particularly high ROI.
In addition, our "Human and AI Collaboration" approach enables zero PoC, allowing you to go live in the shortest possible time.
02
Stable and high quality
Because ABEJA Platform provides modules that have been tested and quality assured, it can always provide a stable development environment from the very beginning.
In addition, it is always equipped with the latest technology, for which R&D is carried out on an ongoing basis. It gives you access to the latest technology and know-how, which means that developed systems and models can always be up to date.
Before systems and models become outdated, you can implement scalable and advanced DX plans that respond quickly and accurately to market changes.
03
Reduced operating costs and labour
Initial investment, equipment costs, operating costs, personnel costs, development costs for the latest modules as needs change and technology evolves...
If you do all the DX themselves to meet your needs, you’ll incur various costs and manpower that can be a heavy burden. ABEJA can take care of this by keeping the modules in ABEJA Platform up to date, including the development and implementation of ML libraries, ML models and AutoML*.
We have a case where ABEJA's "Human-AI Collaboration" approach has enabled implementation in the shortest possible time and reduced post-implementation operating costs by more than 50%.
It can significantly reduce various costs and workloads.
04
Deploying ML models in production based on AutoML is also an option
It's particularly suited to projects where speed and cost are critical.
Based on AutoML*, which automatically handles ML model development tasks and best practices based on diverse data, you can choose advanced operations that can be quickly applied to production, with human support at key points.
AutoML reduces human involvement to the absolute minimum. It delivers dramatic speed improvements and cost reductions.
*AutoML: The machine learning process includes steps such as problem and hypothesis definition, data collection, data processing, feature design, ML model generation and model operation. Among these steps, AutoML automates the steps from data processing to model design.
05
Robust security
It provides a highly secure environment that can be used in industries such as healthcare, finance and local government, where a high level of security is required and where system downtime or privacy risks could lead to major accidents.
In addition, ABEJA does not divert the source code that makes up each module, but customises it for each client company and makes it available to them. You can be sure of information security.
06
Collaborative capability creation
When DX is outsourced, we often hear that the company tends to become dependent on the outsourced company and that knowledge and skills are not built up within the company.
ABEJA provides a wide range of support, including strategy development, business process development, operation and improvement, to meet the needs of the client company with close support. This enables the client company to accumulate knowledge that can become part of its corporate assets, improve its capabilities and build a business model that brings competitive advantage, new results and new opportunities.
How it’s provided
First, we listen to the customer's problems, select the target operation and draw up a design document. Once this has been agreed, the client company supplies the data required to develop the system into ABEJA Platform.
Based on the accumulated data, new data is created and processed in an environment that guarantees computing resources and security, and by combining this data with pre-built AI models, a new system is created to meet the customer needs.
The system is equipped with versatile and flexible modules, not only in the BaaS layer where the AI models are stored, but also in the connectivity, computing, security and core layers, making it possible to build any number of different systems.
If the amount of data required to build a model is insufficient, the human-in-the-loop approach allows even the highly sophisticated AI systems to be built with zero PoC, enabling rapid DX.
Example 1. DX of plant operations at a major chemical manufacturer
Rigorous inspection is essential at a chemical manufacturing plant, where neglecting corroded pipes can lead to serious incidents.
At the client's plant, operations staff took a large number of photographic images of the pipes, and maintenance staff used these images to visually assess the degree of corrosion to determine countermeasures.
Example 2. DX of inventory management operations for specific components at a home appliance manufacturer
