Guide

How to teach a new hire your team's AI workflow on day one

Teach a new hire your AI workflow by having them do one real task with it before lunch on day one. Before they arrive, set up their seat on the company AI plan and their access to the team's shared instructions; in the morning, walk through the one-page AI rules in ten minutes; then they do their first task with the shared instructions, and a buddy reviews the result. By Friday they've done each of their first three tasks the team's way, and you know whether they need help.

Short answer

  • Before day 1: a company AI seat and access to the team's shared instructions.
  • Day 1 morning: the one-page rules in ten minutes, then one real task.
  • A buddy reviews their first three AI-assisted outputs.
  • End of week 1: each of their three core tasks done once, the team's way.

Why day one, not month two

A new hire will use AI from the first week whether you show them how or not; most people already bring their own tools. If nobody shows them, they bring their personal account and their own habits, and you'll spend month two undoing both.

The upside of getting it right is large for newcomers specifically. In the NBER study of a support team, the newest agents improved most with AI help, because it carried what the best agents did. Your shared instructions are that carrier. In the EU there's also a legal reason: Article 4 of the AI Act asks companies that use AI to take measures for their staff's AI literacy, and the European Commission's Q&A says no certificate is needed: an internal record of training is enough.

What to prepare before they arrive

  • A seat on the company plan. ChatGPT Business and Claude Team don't train on your data by default, and the account stays with the company when people leave.
  • Access to the shared instructions for their first three tasks: a shared project, shared skills, or your team's skill library.
  • The one-page AI rules: approved tools, what never gets pasted, who reads AI output before it goes out. See what not to paste.
  • A buddy who does the same tasks and has 20 minutes a day for a week.
  • Three real, low-stakes tasks waiting for them, with past examples.

A worked example

Illustration: a made-up business. Aisha joins the customer success team at a 25-person software company. Her three core tasks are welcome emails to new customers, notes after each customer call, and answers to "how do I export my data?"-type questions.

  • 9:00 Laptop, accounts, and her Claude Team seat, already connected to the team's three shared skills.
  • 10:30 Her manager goes through the one-page rules: customer data only in the company account, never in a personal one; every customer email read before sending; ask in the team channel if unsure.
  • 11:00 First task: she drafts a welcome email for a real new customer by asking Claude for one with yesterday's sign-up details. The shared skill sets the structure and tone. Her buddy, Luis, reviews it and changes one line.
  • Day 2 and 3 Call notes for two calls she sits in on; Luis compares hers with his.
  • Friday Three support answers. Luis finds one mistake: the skill pointed to an old export menu. The skill's owner fixes it for everyone that afternoon.

The last point is the one to copy. A new hire follows instructions literally, so they find the stale parts. Make it easy for them to report what's wrong.

A day-one schedule to copy

  1. First hour: accounts, the company AI seat, the shared instructions visible in their AI.
  2. Ten minutes: the one-page AI rules, read together, questions answered.
  3. Before lunch: one real task with the shared instructions, reviewed by the buddy.
  4. Afternoon: the buddy shows the second task the same way.
  5. End of day: note what was unclear, and send it to whoever owns each instruction.
  6. End of week: each of the three core tasks done once, and a five-minute check-in on what to change.

Freelancers and short contracts

The same day-one plan works for a freelancer, with one difference: they usually arrive on their own AI account, outside your company workspace. How to give contractors your team's AI skills covers how to hand them the same instructions and take access back when the job ends. If you haven't written your shared instructions yet, start with getting your team to use AI the same way, then turn one doc into a Claude skill.

The AI skill below drafts the whole plan around this first day: what must be ready before day one, the first-week schedule, the people to meet, a buddy, a first deliverable, and goals for 30, 60 and 90 days.

Skills for this

All skills

Questions

Should new employees be trained on AI tools during onboarding?

Yes, but on your team's way of using them, not on AI in general: which account to use, what never goes in, and the shared instructions for the tasks they'll do first. In the EU, the AI Act also asks companies that use AI to support their staff's AI literacy.

What should a new hire do with AI on their first day?

Sign in to the company AI account, read the one-page rules, and do one real, low-stakes task with the team's shared instructions, reviewed by a buddy. Doing one task beats reading a guide.

Can a new hire use their personal ChatGPT or Claude account for work?

Avoid it. A personal account may train on their chats by default and leaves with them when they go. Give them a seat on the company plan, or the team's shared instructions on an account your company controls.

AI skill

Have your AI plan a new hire's first 90 days

The Employee onboarding plan (30/60/90) skill has your AI plan the new hire's first 90 days around this first day: accounts, the first week, a buddy, a first deliverable.

Use this skill in your AI

14 days free · No card needed

Sources

  1. European Commission: AI literacy, questions and answers — Article 4 of the AI Act: measures for staff AI literacy, applying since 2 February 2025; internal records are enough (read 2026-11-05)
  2. Brynjolfsson, Li and Raymond: Generative AI at Work (NBER Working Paper 31161) — 34% gain for novice workers; AI passed on experienced workers' practices (read 2026-11-05)
  3. Microsoft and LinkedIn: 2024 Work Trend Index — 78% of AI users bring their own tools (read 2026-11-05)
  4. Anthropic Privacy Center: commercial products — Claude for Work not used for training by default (read 2026-11-05)
  5. ChatGPT pricing — Business: no training on business data by default (read 2026-11-05)