AIToday
Large Language ModelsAI Coding AssistantsAI Business & IndustryITmedia AI+Published: Sep 29, 2026, 06:00 JST

ITR analyst: Japan firms 'almost absent' from harness engineering

ITR analyst: Japan firms 'almost absent' from harness engineering

3 Key Points

  1. What happened

    ITR principal analyst Hiroaki Koumoto said Japanese firms' efforts in harness engineering are 'almost nonexistent.' Only some leading IT vendors and startups use AI agents in software development, he said, while typical companies and SIers are still stuck at an earlier stage.

  2. Why it matters

    Without an integrated agent development process, building a harness has little meaning, so many Japanese firms may keep treating AI as a tool for cutting coding hours rather than a path to better competitiveness.

  3. What to watch

    Handling AI agents safely and securely company-wide, not individual tools like MCP, is the definition Koumoto uses — and he sees a roughly five-year window before the gap with global firms becomes hard to close.

WHO IT HITSThis lands hardest on Japanese business leaders and SIer (system integrator) managers deciding how to spend development budgets and hire engineers. It also matters to IT recruiters and engineering managers who must compete for scarce AI talent against leading IT vendors and startups.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

Koumoto frames the shift as a change in what the industry argues about. Model performance debates have run their course, and the focus is moving to how AI agents are given autonomy — in other words, to harness engineering. Japanese firms, he says, are largely still at the earlier stage of applying AI piecemeal: upstream design, coding, and testing handled as separate tasks, with no integrated agent-driven development process in place.

The deeper cause he cites is structural. Years of 'dependence on major SIers and outsourcing' left many Japanese companies without their own application development know-how or in-house engineers. Business leaders, he says, have started to realize this over the past two to three years, but awareness alone does not create skilled people, and some had hoped AI would let them build applications without them — a hope he describes as an illusion.

His practical advice is to treat application development as the core of business competitiveness and innovation, rather than a cost to be outsourced, and to accept that the first three years may bring no direct return. Koumoto's own comparison is with the shift to agile development after 2015, which took two to three years to take hold in Japan once companies committed. Whether Japanese firms genuinely change course within five years, he suggests, will decide which ones remain competitive.

FAQ
What is harness engineering?
According to ITR's Koumoto, it is not individual tools or communication protocols that support one agent, but a company-wide environment for running AI agents safely and securely so they actually carry out application development.
How far behind are Japanese companies?
Koumoto says harness engineering efforts at typical Japanese firms are 'almost nonexistent.' Only some leading IT vendors and startups are using AI agents in software development, while most companies and SIers remain at the stage before harness engineering.
Why did Japanese firms fall behind?
Koumoto points to years of 'dependence on major SIers and outsourcing.' Many companies lost their own application development know-how and in-house engineers, and some business leaders hoped AI would let them build applications without that knowledge.

Get the latest Large Language Models news every morning

For example, today's edition would include:

  • Meta taps CJ Desai to lead new enterprise AI unitSiliconANGLE AI · 22m ago
  • 1Password ties AI agent access to individual tasksSiliconANGLE AI · 22m ago
  • ServiceNow's Bhakti Pitre: rogue AI agents need risk-based kill switchSiliconANGLE AI · 22m ago

AI-summarized, only the topics you pick: one digest a day via Email, LINE, or Slack.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleMark Penn: AI's ad payoff is new work, not cheaper old work