Itochu Interactive launches training that teaches staff to hand whole tasks to Claude
A Japanese firm is selling companies a course where employees build AI agents for their own work, a sign that the focus is moving from chatting with AI to delegating to it.
The short version
- Itochu Interactive is launching an 'AI agent implementation' course for companies, built on Claude and Claude Code.
- Participants build agents around their own work, not a fixed set of lessons.
- Example tasks: proposal drafts, research reports, minutes processing, data tallying and running a company's own media site.
- What gets built stays with the team, so it does not depend on one person's private prompts.
- Copilot, Gemini and ChatGPT versions are available if a company's security rules require them.
Why it matters
Many workplaces already use AI to draft text, but few have turned that into a repeatable process. This course treats delegation as a skill that has to be designed and taught. It also shows that the hard part is deciding what to hand over, what data to use and where a person checks the result.
What Itochu Interactive is offering
As kn.itmedia.co.jp reported, Itochu Interactive is starting a training program for businesses called the AI agent implementation course. It uses Anthropic's Claude and Claude Code. An AI agent here means a setup that carries out several steps of a job, not just one answer to one question.
The course is not a general tour of generative AI. Each participant picks a real task from their own job and builds an agent for it.
Why a good first draft is not enough
The source uses proposal writing as its example. A finished proposal involves gathering information, organizing data, outlining and then writing. An AI can produce a first draft, but linking it to the earlier steps means spelling out the procedure and the rules for judging each step.
So the key questions are which tasks to hand over, which information to use and at what point a person reviews the work. That is a design problem, not a matter of writing a cleverer prompt.
How the course works
Before training, the provider surveys participants and talks with the company to learn their work problems and AI experience. Examples might include proposal drafts, research reports, running an owned media site, data tallying and meeting minutes. A hands-on session follows, then a review to see whether the work sticks.
Participants write agent skills, which are written-down procedures and rules that Claude can reuse. They also connect agents to internal data and outside tools. The aim is to grow from helping with single tasks to running several steps in sequence.
The results are meant to stay with the team, not sit in one person's private prompts. Sessions can be online, in person or hybrid, and a company can ask for Microsoft Copilot, Gemini or ChatGPT instead of Claude.
What training does not solve
The article is careful to say that knowing how to build an agent does not transform a business by itself. Teams still need to decide who may access what, how confidential information is handled and how outputs are checked.
It also advises weighing the cost of a mistake, not just the time saved, and making sure someone other than the author can maintain each agent. The source gives no results or participant numbers, so how well the course works is still unknown.
What to watch
- Whether Itochu Interactive or its clients share results from the course.
- How companies handle access rights and confidential data once agents connect to internal systems.
Try it
Pick one recurring task and write down its steps, the information it needs, what 'done' looks like and who checks the result. That is the groundwork for handing it to an AI.
Sources and links
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by kn.itmedia.co.jp · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.