Teams usually come to Claude after trying ChatGPT, or because someone on the team started using it and got noticeably better drafts. Either way, the same thing happens next: a few people get very good with it, everyone else dabbles, and nobody agrees on what it's for.

Claude training for teams fixes that by getting everyone working the same way on the same real tasks. I'm an AI implementation consultant with 23 years in marketing, and I built my own AI infrastructure without engineers, so I train on the workflows I run, not on a script.

Start with Anthropic's free courses. I mean that.

Anthropic runs Claude Academy, which its page describes as free courses, tutorials and use cases from Anthropic's education team. One course, AI Fluency: Framework and foundations, teaches a 4D framework of Delegation, Description, Discernment and Diligence across 14 lessons. Another, AI capabilities and limitations, runs to 13 lessons. They're good, they're free, and I'd point anyone on your team who wants the foundations towards them.

So why train with me at all? Because a course teaches one person how Claude works. It can't sit beside your account manager while she turns your messy client notes into a weekly update, or help your team agree what's off limits.

Where a course stops and team training starts

• A course uses example tasks. Team training uses yours: the proposals, briefs, reports and replies your people write every week
• A course teaches individuals. Team training gets everyone working the same way, so good workflows spread instead of living in one person's head
• A course can't set your rules. In a session we agree, in writing, what Claude can and can't be used for in your business
• A course ends when you finish it. I follow up afterwards to see which workflows stuck and fix the ones that didn't

What Anthropic says about your team's data

Anthropic's Privacy Center article on model training, dated 18 August 2026, says that by default it won't use inputs or outputs from its commercial products, such as Claude for Work and the Anthropic API, to train its models. The exceptions are when a user explicitly reports feedback or bugs, for example with the thumbs up or down buttons, or otherwise chooses to allow it. Consumer plans (Free, Pro and Max) are covered by a separate article, and owners of Team or Enterprise plans can switch off the thumbs up and down feedback through the Rate chats setting.

What it means for you: put your team on a business plan, not personal accounts, and decide as a team whether the feedback buttons stay on. If owners want them off, it's one setting, and it's worth doing before the training day rather than after.

What we build in a Claude session

We spend most of the time building, not watching. Depending on your team, that usually includes:

• Claude Projects loaded with your own documents, briefs and style notes, so every chat starts with the right context (my guide to using Claude Projects covers the setup)
• Reusable instructions for the jobs the team does every week, and where it fits, Claude Skills that package a repeatable process
• Workflows for reading and summarising long documents, contracts and research
• Drafting and editing in your voice, with a human check on anything that goes to a customer
• A shared list of prompts and examples the team owns after I've gone

Claude, ChatGPT or both?

I work with both every day and I don't sell either. Plenty of teams end up using both for different jobs, and that's fine as long as the rules are the same across them. If you're still deciding, my Claude vs ChatGPT comparison sets out where each one does better. The training works either way, because it's built on your tasks rather than one tool's features.

The shape of the training day

1. Before the session: we agree the three to five tasks that matter most, and everyone gets access to the same company Claude account
2. In the session: each person builds a Claude workflow on one of their own tasks, with me beside them
3. Before we finish: the team agrees its rules and each person writes down what they'll do differently next week
4. After the session: I check in to see what's working, and the test is simple, is someone doing a specific task differently a month later?

My plain advice for teams starting with Claude

Send your team to Anthropic's free AI Fluency course first, then book hands-on training only once you've picked the tasks you want changed. Put everyone on the same business plan before the day. Decide your rules on customer data before anyone uploads a file. And give one person on the team the job of keeping your shared prompts and Projects tidy, or they'll be out of date within weeks.

Frequently asked questions

Who offers Claude training for teams?

Lilach Bullock, an AI trainer and AI implementation consultant with 23 years in marketing, offers hands-on Claude training for teams, online or in person, built around each team's real weekly tasks. She's independent and works with both Claude and ChatGPT.

Is Anthropic's Claude Academy enough on its own?

It's a strong free starting point for individuals. Team training adds your own tasks, shared rules on safe use and follow-up, which a self-paced course can't.

Does Anthropic train on our team's Claude chats?

Anthropic says that by default it won't use inputs or outputs from commercial products such as Claude for Work to train its models, unless a user reports feedback or chooses to allow it.

Do we need a technical team to use Claude well?

No. Lilach built her own AI infrastructure without engineers, and the training is designed for non-technical teams.

How long is a Claude workshop?

A hands-on workshop runs for 90 minutes, and larger teams can have a longer programme with follow-up.