Hmcomm Inc.

Hmcomm Inc.'s founding traces back to concerns the CEO developed while working as an engineer, project leader, manager, and director at a major SIer (Fuji Soft). Japan has an abundance of outstanding technologies and engineers, yet these are not sufficiently translated into social value or productivity gains for industry — a situation he viewed as a crisis. As cloud services transformed how software is developed and delivered, he came to feel strongly that Japanese engineers' technical capabilities needed to be linked more directly to solving societal problems. Acting on this experience and conviction, he struck out on his own in 2012. He encountered the acoustic analysis technology held by the National Institute of Advanced Industrial Science and Technology (AIST) in Tsukuba and launched the business using the "AIST Technology Transfer Venture Program," a framework for bringing research results into practical social use. Consistent since founding is a passion for sound and the philosophy of "creating value from sound and contributing to society through innovative services." At the time, attention was shifting toward video and image analysis, while few companies were tackling "sound" in earnest — the most technically demanding category of unstructured data. Hmcomm set its technical axis there and has since focused on analyzing and commercializing a broad range of sound data, from human speech to the industrial sounds emitted by factories, equipment, and social infrastructure.

As the company has grown, the talent it attracts has evolved. In the early years, the core was R&D personnel interested in acoustic algorithms, acoustic feature analysis, and machine learning. Later, staff capable of productizing and deploying R&D output joined. Today, the engine of growth is talent with "social implementation capability" — people who not only develop technology but understand customer problems and convert AI into tangible business outcomes.

Hmcomm sees the software development industry facing major structural change driven by the diffusion of generative AI and AI Agents. The conventional model was to hire large numbers of engineers, assemble large development teams, and perform design, programming, testing, and operations manually. Going forward, however, management expects "AI-driven development" — in which AI handles much of the programming and testing — to spread rapidly, with the value of humans writing code declining in relative terms. Accordingly, the billing unit for software development is expected to shift rapidly from person-month-based models to AI Agent licenses, system usage fees, usage-based billing tied to AI compute volume, and ongoing operations/improvement fees. Rather than scaling headcount like a major SIer, Hmcomm intends to leverage its agility as a small, flexible organization, developing and operating a large number of AI Agents to take on a broad range of customer business processes. At the core of this is the Voice AI Agent Platform, "VAAP." Human labor previously delivered as person-month development services will be implemented on VAAP, shifting to a model in which AI Agents are deployed into customer operations. With AI Agents executing tasks just as people do — with outcomes measured and continuously improved — Hmcomm aims to evolve from a one-off AI development firm into a platform company delivering AI Agents on an ongoing basis. That said, it is difficult to design and operate every process with AI alone while also understanding customers' management issues and business requirements. Hmcomm therefore positions the people who face customers directly — from problem discovery through AI adoption, operational implementation, and post-deployment value creation — as "FDE" (Forward Deployed Engineering). FDEs define the customer's operations and target outcomes, build Agents through AI-driven development, and continue to monitor and improve customer operations after deployment. Through this structure, the company aims to build a corporate model in which people and AI Agents coexist and the share of work executed by AI Agents rises continuously.

Defining The Core Battleground

Backed by expanding corporate AI adoption and labor-saving demand in social infrastructure, Hmcomm identifies three core markets with substantial growth potential. First, the market for business process and business model transformation via AI adoption, which will take off in earnest across all industries. Second, the recurring revenue stream from customers' ongoing use of the AI Agents developed and deployed there. Third, the industrial and social infrastructure market addressed by the anomalous sound detection technology "FAST-D." Labor shortages from a shrinking working-age population and aging social infrastructure are structural societal issues, distinct from cyclical swings. Beyond AI substitution for corporate office work, management sees continued expansion in demand for AI-driven labor savings and sophistication in physical operations such as equipment inspection, predictive maintenance, and infrastructure monitoring. Whereas VAAP represents "digital AI Agents" handling in-house business processes, FAST-D is positioned as a "physical AI Agent" handling equipment inspection and infrastructure monitoring. The ability to agentify both the digital and physical domains represents Hmcomm's major medium-to-long-term market opportunity.

Growth Headroom

Even within existing accounts, many business processes remain untouched by AI. Hmcomm has built a track record with major corporates across speech recognition, contact centers, meeting minutes, and anomalous sound detection. Beyond cross-selling existing products, FDEs can uncover new operational pain points within accounts and agentify them, expanding revenue per customer. Rather than selling a single product, the strategy is to deploy Agents/AgentSuites across multiple customer workflows, converting these into recurring revenue via licenses, usage, and maintenance/improvement fees.

Expansion Into Adjacent Markets

The area management sees as offering the greatest growth headroom is FAST-D. Industrial sound and operating data from manufacturing equipment, power, gas, water and sewage, and underground facilities constitute site-specific "dark data" that does not exist on the general internet. Hmcomm has built up a record of real-world deployments analyzing factory and equipment sound to detect anomalies and early signs of failure. Using this accumulated industrial sound data and analytical know-how as a springboard, the company will expand into equipment maintenance and social infrastructure monitoring. Water and sewage inspection, for example, still involves hazardous work and tasks dependent on veteran experience. Going forward, combining FAST-D with inspection robots, drones, and vehicle-based drive-by methods should enable labor savings and greater sophistication in inspection work. Where appropriate, the company will also use alliances and M&A with infrastructure inspection firms, equipment maintenance companies, robot/drone operators, and logistics providers — exploring a business model that goes beyond supplying AI systems to taking on the inspection work itself. While individual deals under consideration are undisclosed, management intends to keep evaluating partnerships and M&A as a medium-to-long-term growth avenue.

Accumulated Speech/Industrial Sound Data And Analytical Technology

Starting out as a technology transfer venture commercializing AIST research, Hmcomm has accumulated technology and know-how for analyzing a wide range of sound data, from human speech to industrial sound. Rapid advances in generative AI have accelerated the pace of AI model development and refinement. At the same time, the importance of high-quality data for training and operational deployment has, if anything, increased. Industrial sound data captured from factories, equipment, and social infrastructure is particularly hard to obtain, and management believes the data and analytical know-how accumulated through years of customer engagement and real-world deployment constitute a competitive advantage.

End-To-End Social Implementation Capability

Supplying AI component technologies or middleware alone does not deliver business transformation for customers. Actual deployment requires understanding the customer's workflows, core systems, security requirements, and operating structure, then embedding AI into existing operations. Hmcomm has spent years embedding AI into customers' live operations — contact centers, anomalous sound detection systems on manufacturing floors, and more. The ability to deliver end-to-end — from problem discovery and AI model development to integration with legacy systems, production rollout, and post-deployment operations and improvement — differentiates the company from vendors focused on supplying component technology.

On-Premise And Closed-Network Support Via Proprietary AI Engines

Hmcomm builds on its own AI engines while also incorporating open-source technology, strengthening a technology stack it can manage and deliver within its own environment. This enables deployment into on-premise environments, closed networks, and edge environments — settings that cloud-based external AI services alone struggle to serve. Companies handling confidential information, as well as the public sector, financial services, manufacturing sites, and social infrastructure, require AI deployments that account for security, communications environments, and running costs. The ability to control AI models, data, and processing environments in-house — rather than simply building applications on external generative AI APIs — and to guarantee quality and security tailored to customer requirements is a key competitive advantage.

Alongside conventional revenue and account metrics, the company will emphasize KPIs that track the shift from a person-month business to an AI Agent recurring business.

Historically, AI Products revenue has included NRE (initial implementation development) alongside recurring revenue such as monthly licenses, making revenue sensitive to the timing of large deals. Going forward, the company plans to separate NRE from recurring revenue where feasible and to progressively disclose ARR/MRR, contracted company counts, and agent counts, giving investors a clearer view of the shift toward a recurring business.

Management frames revenue growth in three stages. Stage 1: FDEs identify the customer's management and operational challenges and define the target processes for AI adoption along with the outcomes to be achieved. Stage 2: Design and build the Agents/AgentSuites required to solve those challenges and connect them to the customer's core systems and workflows. Stage 3: Run those Agents continuously on VAAP, accumulating recurring revenue from monthly licenses, usage, and maintenance/improvement.

Once this cycle is established, more FDE engagements lead to more Agents in production, driving a build-up in ARR and usage revenue. This creates a repeatable growth model in which not only revenue but also gross profit and operating income expand continuously. Combining cross-selling into existing accounts, expansion of core products and services, ongoing Agent delivery via VAAP, and capital and business alliances, the company targets sustained growth of roughly 30% per year over the medium-to-long term. On inorganic growth, M&A is positioned not as a simple means of adding revenue scale but as a lever to accelerate the VAAP business and organic growth. In particular, by converting the customer relationships and talent held by software and SES companies into FDE and AI-driven development, the company aims to shift the structure from person-month business to AI Agent business. Personnel joining through M&A will be reskilled into FDEs, broadening capacity to discover customer problems, while the AI-driven development platform "FD-X" raises development productivity for Agents/AgentSuites. Management notes this could lead to an "Agent Factory" concept in which FDE deal wins, Agent development, production rollout, and ARR conversion scale in sequence — though at this stage it remains under continued study.

There are two main paths to improvement.

First, development-side efficiency through FD-X. Programming, testing, and documentation will be AI-driven, building a framework for developing and improving Agents efficiently and semi-automatically. The goal is a structure in which the same headcount can deliver more Agents/AgentSuites to customers.

Second, SG&A efficiency through the "AI Head Office" concept. Routine tasks currently performed by staff in accounting, HR, general affairs, legal, and sales administration will be agentified, creating an organizational structure in which SG&A does not rise in proportion to revenue. By using AI to improve efficiency in both development costs and SG&A while raising the recurring revenue mix and gross margin, the company targets a consolidated Operating Income Margin of 20% or higher over the medium-to-long term.

On M&A, the focus is not on consolidated revenue scale alone but on deals that connect the target's customer base, talent, technology, and data to VAAP, FDE, and FD-X, accelerating group-wide organic growth. Post-acquisition, the company will prioritize synergy creation with existing businesses and profitability improvement, driving PMI with capital efficiency in mind.

ROE / ROIC Targets: No specific numerical targets have been announced at present. Going forward, the company intends to consider setting and disclosing capital efficiency metrics, balancing the returns on growth investment, cost of capital, and margins.

The largest downside risk, in management's view, is a delay in the business model transition to VAAP. Hmcomm has technological strength and a deployment track record in voice AI and anomalous sound detection, but recognition as an integrated AI Agent company still needs to be built further. VAAP therefore needs standardized Agents/AgentSuites that let customers verify benefits within a short timeframe, demonstrating shorter implementation periods and overwhelming price advantage.

The second risk is insufficient conversion of FDE engagements into ARR. If FDE engagements end as one-off AI solution development projects, revenue may grow but the targeted recurring revenue structure will not materialize. Management therefore designs the target processes and the post-deployment license/usage model from the outset of each FDE engagement, and will closely manage the conversion rate from FDE to Agents/AgentSuites and on to ARR.

The third risk is delayed PMI and talent conversion following M&A. If acquired companies simply continue their traditional person-month business, the group's earnings structure will not change materially. The policy is to provide on-the-job training in FDE, FD-X, and VAAP from an early stage post-acquisition, place and reskill personnel according to aptitude, and manage PMI progress quantitatively.

Executive Compensation

At present, there is no share-price-linked compensation or stock option scheme. However, recognizing that corporate value is linked not only to revenue but also to growth rate, ARR, and margins, management operates against internal business plans set above the disclosed guidance and aims to clear them. The company notes it will need to consider a compensation structure linked to these metrics going forward.

Succession Planning

The company operates a company system in which each division is treated as a standalone business, with executive officers serving as company heads responsible for people, assets, and capital. They are held accountable not only for revenue but also for profit and capital efficiency, and a pathway has been established whereby those who deliver results advance from executive officer to director, and then to titled director with a defined mission. In addition, one of the evaluation criteria for each company unit is to "designate a successor and spin the unit off to that successor within two years," creating a mechanism whereby company units multiply and future board candidates are cultivated. As for the CEO himself, he acknowledges that if a capable successor emerges — depending on performance — he would hand over the baton and take on a different role.

Employee retention has improved markedly versus the past and has reached a reasonable level (no specific figures disclosed). Although the industry as a whole has high mobility and intense churn of top talent, management attributes improved retention to the opportunity to work on cutting-edge initiatives and to a problem-solving approach grounded in understanding customer challenges.

Behind this lies a change in the industry structure itself. The traditional division of labor across PM, SE, programmer, and tester roles has changed dramatically, and management sees the role of talent itself at an inflection point. The most important theme, therefore, is how to shift these personnel toward higher-value-added work and transform their roles.

Beyond its own employees, the company intends to share its definition of FDE broadly with group companies and future acquisitions, driving a human-capital repositioning of the business. Management states that its responsibility is to create a state in which it can continue solving societal problems through software — in other words, a state in which its people can earn a living over the long term.