Onboarding someone into a company that uses AI
August is a hiring month. Two paragraphs in your welcome pack will save a year of unclear expectations.
Somebody starts with you in August or September. Their first week includes a laptop, a password policy, a fire exit briefing and a folder of documents. Almost nowhere does it include a paragraph about AI, and it should.
Not because new staff are reckless, but because in the absence of a stated rule they will apply the one from their previous job, and you do not know what that was.
What the paragraph needs to say
Three things, and nothing else.
First: which tools are approved here, by name. Second: what must never go into them — give two or three concrete examples from your business, not a category like « confidential data ». Third: who to ask. A name, not a mailbox.
If you wrote the one-page policy this blog suggested on 21 January, this paragraph is a summary of it plus a link. If you did not, writing this paragraph is the easiest way to end up with the policy.
The part everyone forgets
Say whether using AI is expected, permitted, or tolerated. These are very different, and new employees guess badly.
A junior who assumes it is tolerated will hide the fact that a draft was machine-assisted, which is exactly the outcome you do not want. A junior who assumes it is expected will use it for a task where your firm’s value is that a human did the thinking.
One sentence resolves this. Most companies have never said it out loud, including to the people who have been there ten years.
Do it for the existing staff too
The new-joiner pack is a convenient excuse. Whatever you write for them should be sent to everyone, once, with a subject line that says it is a two-minute read.
The 14 January post argued that the most common AI problem in a Swiss SME is not misuse but silence — people using tools nobody discusses. Seven months later, in most companies, that is still the situation. August is a quiet month. It is a good month to end it.