日本語A&M GROUP Official SiteProfessional AI Multi-AgentSound Engine DOMEarEffect

Handling multiple tasks at the same time, an AI system endowed with keen senses and broad, meticulous thinking

Concept

Handling multiple operations at the same timeAn AI system with sharp five senses and broad, meticulous thinking

Perception AI that dramatically transforms sound, and Professional AI that combines human sensibility and thinking. An international patent for Sound Engine DOM, and an AI multi-agent that drafts professionals' reports and revises them itself.

Concept

Forward Deployed Engineering

The same engineer is in charge from the interviews about your work through to implementation

We go on site, listen to how the work is done, and design and run AI systems on the spot. Using AI agents, AI orchestration, and AI multi-agent systems as appropriate, we operate 8 large-scale AI systems of our own in real business, including a multi-agent system that oversees the operation of the entire business, and these systems handle the work of several thousand people.

Technology Details

AI Consulting

We analyze your work and design it into a form that AI can process

Much of our system development begins with business analysis in consulting. Professional AI also grew out of the analysis and prototyping of professional work.

Consulting Details

AI Multi-agent

Multiple AIs with different roles share the work, and a central AI brings the whole together.

The multi-agent system that oversees the operation of the entire business, consolidating the current status, matters requiring management decisions, and issues across all businesses, runs with 7 agents.

Architecture Details

Professional AI

Professional AI: PoC succeeded, full-scale productization begins, and contract development also starts

We draft everything from audit working papers to audit reports in the prescribed format. We plan to offer this to professionals in the audit field.

Product Details

Product — Sound Engine

Sound Engine DOM, which changes sound quality through software

A sound engine that rebuilds the audio signal sent to a device in software and changes the tone as much as if the hardware had been replaced. A next-generation Perception AI that manages sound quality in software without replacing equipment. In the broadcast field, we delivered audio processing software after verification in the field.

DOM Details

Distribution

We import and sell 9 brands from Europe and the US in Japan

IGS Audio, Cordial, hama, EQUI=TECH, SKY AUDIO, SessionDesk, Hum Audio Devices, Marian, and DURABLE.

Exclusive Import Distribution Details

Concept

Handling multiple tasks at the same time

An AI system with keen senses and a bird's-eye, meticulous way of thinking

— Perception AI that dramatically transforms sound, and Professional AI that combines human sensibility and thinking —

Perception AIAI that takes on the role of human hearing

Perception AI

This is a next-generation Perception AI that handles human hearing and controls sound quality. Sound Engine DOM rebuilds, in software, the audio signal sent to a device and substantially changes its tone. The change is an improvement in sound quality comparable to replacing the hardware. Sound quality is managed in software, without replacing equipment. It runs built into broadcast transmission systems, streaming infrastructure and handheld devices (EarEffect), and within the multi-agent system it works as the agent responsible for "hearing."

Learn moreClose
  1. Receives the audio signal
  2. Controlling acoustic characteristics
  3. Delivering to equipment in every industry
  4. To the listener's ears
  • Built into broadcast playout systemsDeliveredAudio processing software delivered after listening tests and verification at broadcast sites.
  • Built into distribution systemsTrial deployment decidedIn broadcasting and streaming, a new trial deployment was decided following a demo.
  • Listen on your own device (EarEffect)AvailableWindows and Android Auto versions released, with a multilingual interface. 140,000 downloads worldwide in total.
  • International co-development and an international patentCo-developed with members from the US, UK, Germany and Sweden (since 2020). The method is covered by Patent No. 7251007 (Japan, US, Germany, UK).
No. 7251007
International patents for DOM (Japan, the US, Germany and the UK)
140,000DL
EarEffect global cumulative downloads
4 countries
International joint development (United States, United Kingdom, Germany, Sweden)

It can be embedded in equipment and services in every industry where sound is played, from broadcast and streaming to automotive, rail and bus, aviation, space, film and video, games, TV and home appliances, medical, telecommunications and calls, and AI voice.

Sound Engine DOM

Professional AIAI that takes on the writing work of professionals

Professional AI

This is AI for professionals who produce many reports. It runs on a combination of orchestration and multi-agent design. A coordinating agent (orchestration) assigns the work, and multiple agents with roles such as reading, cross-checking, drafting, and sourcing (multi-agent) share the tasks and work together. They read across materials, draft in the prescribed format, and keep the supporting evidence. Final approval is made through review by the professional.

Learn moreClose
  1. Set external information
  2. Learn the writer
  3. Write in a distinctive style
  4. Corrects itself
  • Self-correctionChecks drafted text against the source materials and corrects any discrepancies on its own.
  • DetectionDetects and flags unsupported statements, formatting defects and swapped source documents.
  • Collecting external articles in real timeCollects paid articles from newspapers and news agencies (e.g., Nikkei, The New York Times, Kyodo News) and reflects them in each report with sources cited.
  • Reflecting the writer's individualityFor each user, past writing, reports and personal profile information are loaded so that working papers and reports reflect the writer's individuality.
159 documents
Documents ingested for a PoC project
In operation
In operation in the audit field, toward productization
approx. 6,000 characters
Text drafted from a single form

It can be used by professionals who produce large-scale reports, such as audit departments of listed companies, audit firms, lawyers, judicial scriveners, tax accountants and researchers.

Professional AI

ExpertiseAreas of expertise

Consulting

Consulting

Not limited to AI, we determine what to build to improve your operations, whether business systems, web systems, or apps, and confirm the requirements in writing. If off-the-shelf tools are enough, we tell you how to use them. Much of our development starts here.

Assess
We interview you about your operations and decision criteria, then judge whether to solve the problem with AI, a business system, a website, or an app, or whether nothing needs to be built at all.
Putting it in writing
From the questionnaire and materials, we finalize in a document how your operations actually run and the requirements, and compile an assessment of whether they can be realized.
Test and verify
We verify effectiveness and accuracy with a prototype on real data, and you decide whether to proceed to development, with the evidence to support that decision.
Choose the form your inquiry takes
Choose the form of engagement that suits your stage, from consultation, advisory and hands-on support. Costs are presented through the AI quotation.
Details on Consulting
Forward Deployed Engineering

FDE (development embedded on site)

The same engineer handles everything from interviews and requirements definition through design, implementation, on-site verification, and operations. Rare professionals, of whom few are qualified, are on staff as our engineers. With no one placed between requirements and implementation, and a structure built around our in-house development AI agents, we can increase the number of times we rebuild in response to feedback from the field without being constrained by man-hours.

There are no handovers between staff members.
The engineer who did the interviewing designs and implements the system directly. There are no requirements misunderstood in handoffs between stages, and no waiting time.
Rebuilding is fast
With a structure built around our in-house development AI agents, you can increase the number of times you rework a product based on feedback from the field, without being limited by labor hours.
We verify before delivery
We compare AI and human judgments using real data and report even the reasons for any discrepancies before moving to operation in actual business.
We continue operation and improvement after delivery.
Drawing on our experience of operating 8 AI systems that do the work of several thousand people in-house, including the multi-agent system that oversees the operation of the entire business, we continue to make improvements after delivery.
Details on FDE
AI & Systems

AI and System Development

In our own company as well, we run 8 large-scale AI systems in actual operations, each taking on the work of several thousand people, including a multi-agent that oversees the operation of the entire business. AI drives our daily work: selecting government tender opportunities and drafting documents, Amazon orders, inventory, advertising and finance, shipments and inventory shared with our logistics warehouse, drafting audit working papers, management oversight, managing the audio app business, sales, and even updating this site.

AI systems
We choose among AI agents, AI orchestration, and AI multi-agent setups to suit the work.
Web and Business Systems
We build systems for orders, inventory, billing, and administration, as well as web systems for applications, reservations, and memberships, and we repair existing ones.
Apps
We handle apps for Windows, Mac, Android and iPhone, from planning through release.
Multitasking AI
We bind multiple AIs into a single system and run several operations at the same time.
Details on AI and Systems Development

Architect's Note

Ideas that do not yet exist in this world: I love the work of devising entirely new AI mechanisms and structures.

Representative Director, President, CEO and CTOHirotoshi Sakamoto
CEO MessageCEO Message Our Representative, Hirotoshi Sakamoto, explains our thinking and the company we aim to be.Read the message
Patent
No. 7251007
Sound Engine DOM Patents (granted in Japan, the United States, Germany, and the United Kingdom)
AI agents
180 agents
Number of AI agents we have developed
Capacity
2,000 person-equivalents per day
The human workload handled by the AI multi-agent system that we operate in-house
Multi-agents
8 systems
Number of AI multi-agent systems operating in our internal operations
Records
6,137,685items
Total number of information items managed collectively by AI (including nationwide corporate registration records)

Scale — The Size of the Work Handled by AI

The volume of work handled every day by AI systems that do the work of several thousand people, including the multi-agent system that oversees the operation of the entire business

AIs with different roles read information, make judgments, and keep records at set times. Checking public information alone, such as public notices from government agencies, comes to approx. 100,000 items every day. 8 AI systems process, day after day, a volume that would require several departments if done by hand.

Workload · from system logsValues taken from each system's records (as of November 9, 2026)

Collecting and assessing information

Public data read per day

approx. 100,000cases/day

Checks bid notices published by government agencies and others 4 times every day and judges deadlines, participation conditions, and fit with our company

Tracking projects

Cases tracked

1,108cases

Cases AI keeps tracking. Cases passed over are also recorded with the reasons

Core business systems

Items under one roof

4,084items

Handle orders, inventory, pricing, advertising and finance on a single platform

Logistics

Units managed

3,900units

From inventory allocation to shipping instructions

Overall supervision

Agents in the business-wide multi-agent system

7agents

Consolidating the current status, matters requiring management decisions, and issues across all businesses, and linking them to improvement

Connection

Tools wired to AI

86tools

The functions of each system can be used directly from an AI assistant

"Several thousand people" is a value we calculated ourselves: we converted the workload above into the effort it would take people to process it with the same tasks, and expressed it as the number of people needed to finish it in twenty-four hours. It is not a measurement by a third party.

ArchitectureChoosing how to build AI to suit each business task

AI Agent

AI ⟶ { tools · data · procedures } ⟶ done

AI agents

It takes on a defined role, switches between tools and data as needed, and advances the procedure on its own.

Our example

  • AI agents that handle our in-house development
  • Import the table, read the meaning of each column, and make the determination
  • Browser automation: filling in forms and collecting information

AI Orchestration

Input ⟶ [ Central control ] ⟶ AI ⟶ AI ⟶ Output

AI orchestration

A central mechanism controls AI, external services, and scheduled processes, and runs them in the order you define. If a process stops, it is detected and the person in charge is notified.

Our example

  • Core system for e-commerce businesses: connecting orders, inventory, prices, advertising, and finance
  • AI system for government bidding: scheduled retrieval of public notices, deadline judgment, and submission records
  • Scheduled collection, inspection and notification
  • A connection mechanism that lets an AI assistant read and write information in each system

AI Multi-agent

{ Read · Collect · Inspect · Evidence · Draft } ⇄ Central ⟶ Human confirmation

AI multi-agent

Multiple AIs with different roles share the work, and a central AI brings the results together and finalizes them. A person makes the final decision.

Our example

  • Professional AI that drafts audit working papers in the prescribed format
  • AI system for government bidding: from collecting public notices to drafting proposals
  • A multi-agent system overseeing the operation of the entire business: 7 agents
  • Drafting improvement proposals and applying them through approval

Differences in architecture (structure) explained in detail

Systems in productionSystems that combine multi-agent and orchestration approaches and are actually in operation in our own business

A multi-agent system that oversees the operation of the entire business

Outer ring — AI Multi-Agent

Outer ring: systems with different roles

Each system handles its own operations, while a central multi-agent that oversees the operation of the business as a whole consolidates the current status of each business, matters requiring management decisions, and issues.

  • AI system for government bidding
  • Core system for e-commerce businesses
  • Shipping and inventory management system
  • Audit working paper system
  • AI agents for development
  • Access points for AI assistants

Inner ring — AI Orchestration

Inner circle: processing flow

Central control runs the flow in a set order, from the scheduled start through collection, checking, proposal, approval, execution and verification. Steps that require judgment include human approval.

  1. Scheduled start-up
  2. Collection
  3. Inspection
  4. Proposal
  5. Approval (human)
  6. Execution
  7. Verification

This website

This site itself is
a working example of AI use.

This site is designed as an AI web system. It is a place where you can see AI multi-agent orchestration, which processes many complex tasks in parallel, in actual operation and deepen your understanding of it. At the same time, it is operated as a working example to help you consider concretely how to advance your own company's use of AI.

Business AI multi-agentHandles the work of 2,000 people per dayBiddingOrders & inventoryShippingAudit papersFinanceBusiness oversight AIGathers information across the businessSite oversight AIHandles the work of 72 people per dayEditing & updatesEnglish translationSearch optimization (self-directed)QuotationsReception & registration checksAI securityManager's approval & sign-offSEO proposals & final sign-off on quotesApplied to all pagesWebsite of 112 pagesLearns from search data and inquiries for the next updateStrategic thinkingThinking AI that supports management decisionsOrganizes issues, options and rationaleRequests a decisionManagement Decisions (CEO)Policy, approval and final decisionsRecords decisions and makes them rules for all AIReports results and insightsPolicy and rules go to all AI
  • AI web systemDesigned so that AI handles everything from reception, corporate registry checks, interviews and quotations to search optimization and updates.
  • AI multi-agent orchestrationA supervising AI directs AIs with different roles, processing many tasks in parallel.
  • A working example of AI useUsing a system that is in operation, we show concretely how to build AI into your business operations.
How this site is operated

FDEForward Deployed Engineering

The same engineer handles everything from interviews and requirements definition through design, implementation, on-site verification, and operations. In typical contracted development, different people are in charge of interviews, requirements definition, design, implementation, and verification, and every handoff between stages leads to misunderstood requirements and waiting time. Because the same engineer sees the work through from start to finish, the requirements you share with us become the system that goes into operation. Rare professionals, of whom few are qualified, are on staff as our engineers.

  1. Consultation

    We go on site and ask about the operations and the criteria for decisions.

  2. Requirements Definition

    We document the requirements through a questionnaire and materials. We have a track record of organizing prescribed forms by type and finalizing in writing, through a questionnaire, how operations actually run and the requirements.

  3. Design and Implementation

    We select the configuration from AI agents, orchestration, and multi-agent setups, and implement it.

  4. Verification

    AI and human judgments are reconciled on real data, and the reasons for any mismatches are reported.

  5. Operation & Improvement

    We operate it in actual business, detect faults, and keep improving it.

Short timeframeFrom requirements definition to on-site operation
Every case cross-checkedThe scope in which AI judgments were checked against staff judgments
8Systems operated in our own business
ProductsInquiries About Adopting or Trying ProductsInquire here DevelopmentRequest System DevelopmentMake a request

Case study

Case Study

We present the flow and the finished form in the form of reports. Each one states clearly whether it is based on our actual operations or is an example assumed from how such work is normally done.

We review public notices every day and select the projects to bid on.
Example based on our own operating experience

AI system for government bidding

Checks approx. 100,000 notices every day
and selects the cases to bid on.

This is our own example. Simply tracking the notices from each agency took people the entire day, leaving no time to refine the substance of proposals.

  1. Collect public notices at set times and assess eligibility, deadlines and fit
  2. Link amendment notices and duplicates to the original case
  3. We select the target projects and report to the executives
  4. Draft the proposal and submission documents, and cross-check them

ReportReport for this case

Flow
AI handles collection → assessment → selection → drafting → cross-checking, while executives make the decisions and submit.
Deliverables
  • Management list of 1,108 cases
  • Drafts of proposals and submission documents
  • Results of the pre-submission check
Points to confirm
  • Are any cases past their deadline still outstanding?
  • Whether executives make the final decision on whether to bid
  • Are retrieval failures being detected?
We manage everything from orders to finance for four thousand eighty-four items as a single set of data.
Example based on our own operating experience

Core system for e-commerce operations

Everything from orders to finance for 4,084 items
is managed as a single set of data.

Here is an example from our own operations. The figures for Amazon 1P (Vendor) and 3P (Seller) were on separate screens, and we pieced together profitability and inventory for each product by hand.

  1. Connect orders, inventory, pricing, advertising and finance by product
  2. Detect unprofitable items and stockouts
  3. Work with the shipping and inventory management system and confirm the process through to delivery
  4. Make it possible to check the status from an AI assistant

ReportReport for this case

Flow
Run one piece of data through detection, instruction, and delivery, every day.
Deliverables
  • A product-by-product list of profitability and inventory
  • Abnormality detection and notification
  • Inventory allocation and shipping instructions for 3,900 units
Points to confirm
  • Is a wrong decision to stop being made because data is missing?
  • Are sales models (1P (Vendor) / 3P (Seller)) distinguished and handled separately?
  • Are delivery delays visible?
AI drafts reports in the prescribed format from your materials and corrects its own errors.
Example based on our own operating experience

Professional AI

The AI drafts a report in the prescribed format from source materials
and corrects its own errors.

This is an example of our products. In a professional organization, a great deal of time was spent on reading more than 100 documents for a single case and producing documents in the prescribed format.

  1. Import the materials and manage them as evidence
  2. Divide forms by type and draft them
  3. Check against the materials and correct itself
  4. A professional reviews and finalizes

ReportReport for this case

Flow
Run this cycle for each case: intake → drafting → self-check → review and finalization. The PoC succeeded in auditing for the first field.
Deliverables
  • Drafting in the prescribed format (clearly marked as an "AI draft")
  • Linking to source documents
  • Version comparison and review records
Points to confirm
  • Whether any statements without supporting grounds remain
  • Is it getting closer to the writer's text?
  • Whether finalization is done by the professional

Contact

We take inquiries about adopting products and developing systems.

Contact Us