Claude for Teams: How to Roll It Out and Get It Used

Cover card with the title Claude for Teams: How to Roll It Out and Get It Used, in white text on a dark background with a cyan glow

Claude for teams means giving a group shared Claude seats, shared projects and one set of rules. Buy the Team plan, pilot one group on one repeat task, write a one-page data rule, keep your best skills in one shared folder, and measure weekly use.

If you want someone to set it up with your people in a room, I teach this in hands-on AI workshops. Below: what the Team plan includes, a four-week rollout, the rules to write first, and how to tell it is working.

What does Claude for teams mean?

It is a way of working, and also the name of a plan. The way of working: instead of ten people each trying Claude alone and getting ten different results, the team shares its setup, its instructions and its rules. The plan: Anthropic’s Team plan, which is built for that.

On its Team plan page, read November 20, 2026, Anthropic says projects and knowledge bases are shared, so one member’s setup becomes the team’s starting point. It lists single sign-on, domain capture, spend caps at the organization and user level, and connectors to tools such as Gmail, Google Drive, GitHub, Slack and Microsoft 365. It also says there is no training on your data by default.

If you run a company with only a handful of people, the owner’s version of this story is in Claude for small business. This post is for the next step: you have a team, several people will use Claude on real work, and you want it to stick.

Which Claude plan does a team need?

Two or more people on shared work should look at Team. One person can stay on Pro. Plans as listed on Anthropic’s pricing page, read on November 20, 2026:

Plan Who it is for Price shown on that date Notes from the page
Pro One person doing everyday work $20 a month billed monthly, or $17 billed annually Training on your content is a setting you control
Team, standard seat Teams of 2 to 150 seats $25 per seat a month billed monthly, or $20 billed annually More usage than Pro, shared projects, central billing, SSO, admin controls for connectors, no model training by default
Team, premium seat Heavy users on a team $125 per seat a month billed monthly, or $100 billed annually 5x the usage of a standard seat
Enterprise Larger companies See the pricing page Everything in Team plus role-based access, SCIM, audit logs, a Compliance API, custom data retention controls and IP allowlisting; Anthropic lists a 20-seat minimum

The pricing page was long and parts were cut off when I read it, so confirm the details for the plan you plan to buy. Seat names and prices have changed more than once this year.

Two practical rules. Put most people on standard seats and only upgrade the ones who keep hitting the limit. If you start running out of usage on the entry plan, that is a great problem, because it means you are really using it, and that is when to move that person up.

How do you roll Claude out to a team?

Start smaller than feels exciting. Anthropic’s own administrator guide in Claude Academy, which shows no publish date, recommends a pilot group first and 2 to 4 weeks of watching adoption before expanding in stages. Its example pilot is 50 to 100 users, which is for a large company, so scale it down: for a team of ten, a pilot is three or four people.

  1. Pick one owner. One person answers questions, checks the output and decides when to widen. A rollout nobody owns turns into ten private experiments.
  2. Pick one task and today’s number. For example: “Writing the weekly client update takes four people two hours each.”
  3. Choose two or three champions. Anthropic suggests 2 to 3 champions per department or team, with early access and a place for peer questions. These are the curious people, not the senior people.
  4. Set the plan and seats. Team for a group, standard seats to begin.
  5. Write the one-page data rule (next section) before anyone pastes customer data.
  6. Set each person up with context. When you start using Claude, it has no idea who you are, what your company does, what rules it should follow, or where to put the files it creates. A short note about the business, the tone and the rules fixes most of that.
  7. Run it four weeks in approval mode, then compare to the baseline. Widen, change or stop, based on hours and errors.
Free resource: Claude Setup template Free resource · Template Claude Setup template A starter setup with a folder structure, memory files, global rules, favorite skills and safety hooks, so Claude knows who each person is before they ask it anything. Get it free →

When I start with a new client, the first thing I ask is how good they are with Claude and other AI tools, whether they use skills, scheduled tasks, hooks or CLAUDE.md files, and I ask them to be honest so I can guide them at the right level. On a team that matters more than it sounds, because one person is often far ahead of the rest, and a rollout pitched at the wrong level loses both groups.

What should a team use Claude for first?

Pick work that repeats every week, where a wrong answer is easy to spot. My goal is really small, easy, quick wins. When a team sees how fast something gets built, they say wow, it was that easy, and they want to automate everything else.

Team A good first task What a person checks before it goes out
Sales First-draft follow-ups after a call Names, prices and anything promised
Marketing First drafts from a brief and past examples Facts, claims and tone
Operations Weekly status report from the numbers you already track Two or three totals against the source
Support Suggested replies to common questions Policy, refunds and anything legal
HR Job descriptions and interview guides Fairness, pay language and local rules

Not every problem is an AI project. Sometimes it is just an inventory system, and even a great system is pointless if people will not use it. If the task is broken, fix the process first.

What rules should a team set before it starts?

Write one page. It should say which data may go into Claude, which may not, which connected tools it may read, and who approves what. For the longer version, see AI governance.

Two things I would put on that page:

  • Read yes, write and delete by approval. I always give Claude permission to read, because that does no harm, but I never give it blanket permission to edit or delete. When it wants to write, edit or delete, it asks and I approve. There are horror stories of an AI that deleted someone’s files. For a team, that is a setting in each person’s tools, and it should be the default.
  • Do not make the rule only a list of no’s. An AI usage policy that only limits people makes them afraid to do anything with AI. I want people to feel free to use it for everything, inside clear lines on data and approvals.

For admin controls, Anthropic’s Team page lists single sign-on, domain capture and spend caps, and its administrator guide says to complete a security review with your IT and security people. Domain claiming is optional and cannot be undone, according to that guide, so decide it on purpose.

How does a team share what works?

Put it in writing, in one place, and let people improve it. Three things spread well:

  • Projects. On Team, projects and knowledge bases are shared. Make one per repeat task, with the instructions and examples inside.
  • Skills. A skill is just a set of instructions for how you like to do something. Once a task works, save it as a skill so Claude does it the same way every time. If you are using Claude without skills, you might as well use ChatGPT: a long chat gets summarized and your original instructions fade, and a skill keeps them.
  • A shared folder. It is fine to install skills into Claude, but I keep mine locally in folders instead. That way I can edit them faster without repackaging and reinstalling, share them with my team, and back them up with Dropbox or Google Drive.

Skills are where teams lose the most. Every time I speak to a room of people and ask who uses Claude, everyone raises a hand. When I ask who uses skills, almost no one does, and skills are so important.

Video: If You're Not Using Skills, You're Using Claude Wrong Video · 15 min If You're Not Using Skills, You're Using Claude Wrong I show how to install a skill, build your own and edit it so it gets better every time you use it, which is the step that turns one person's chat into something a whole team can repeat. Watch the video →

How do you know Claude is working for your team?

Count use and time, not opinions. Anthropic’s administrator guide lists targets for large rollouts. Treat them as a yardstick to scale down, not a promise:

What to track Anthropic’s target in its guide
Weekly active users 70% of licensed seats
Messages per user per week 25 or more
Projects and Artifacts in use 40% of active users
Time saved per user per week 3 or more hours
User satisfaction 4.0 or higher out of 5

The guide also suggests a rhythm: a short message before launch, tips in week one, weekly tips afterward, and department follow-ups 2 to 4 weeks after training. Those are Anthropic’s suggestions for companies at its customers’ scale. For a team of ten, a 15-minute Friday check-in does the same job: what did you try, what worked, what do we save as a skill.

If weekly use drops after week two, the usual cause is not the tool. It is that nobody owns it, or that the task was never worth doing.

Where do team rollouts go wrong?

Three places show up again and again.

  • Fear. I have said for years that the most common barrier to automation is employees’ fear of change and worry about job security. Support them as they adapt, and involve them in the decision so they feel invested in the company’s success. Say out loud what the goal is, and ask the team what to automate first.
  • Accounts without a use case. Handing out seats and hoping is the version that fails. Name the task before you name the seats.
  • No review. A workflow no one checks is just a faster way to make errors. Name one person who looks at the output each week.

Does your team need training?

Usually yes, and it should be short, practical and built on the team’s real work. I run real workshops where people get work done on the spot, in small groups, one to ten minutes at a time, and finish by getting everyone to write down and commit to one action. My Claude workshop needs a minimum of an hour and a half, because we set Claude up completely on each person’s computer.

For a longer plan, see AI training for employees, and for the basics of building a shared understanding, AI literacy. If you manage the team yourself, AI for managers covers your side of it.

What are the next steps?

  1. Name one owner and two or three champions.
  2. Pick one repeat task and write down today’s number.
  3. Choose Team and begin on standard seats.
  4. Write the one-page data rule before the first customer record goes in.
  5. Run four weeks, then save what worked as a shared skill or project.

If you want your team to learn Claude on a real task from their own week, see my AI workshops.

Updated November 20, 2026. Sources opened on that date: Anthropic’s Claude pricing page, its Team plan page, and the Claude Academy Enterprise administrator guide.

FAQs

What is Claude for teams?

It is using Claude with shared seats, shared projects and one set of rules instead of everyone experimenting alone. Anthropic sells it as the Team plan, which covers 2 to 150 seats and adds central billing, single sign-on and admin controls.

How much does the Claude Team plan cost?

On November 20, 2026, Anthropic's pricing page listed a standard seat at $20 per seat per month billed annually or $25 billed monthly, and a premium seat at $100 billed annually or $125 billed monthly. Check the page the day you buy, because plans change.

How many people do you need for the Claude Team plan?

Anthropic's pricing page lists Team as for teams of 2 to 150 seats. Above that, look at Enterprise, which Anthropic lists with a 20-seat minimum.

Does Claude train on your team's data?

Anthropic lists model training as none by default on Team. On Pro and Max it is a setting each person controls, which is one more reason a team should not run on personal plans.

How do you roll Claude out to a team?

Pick one pilot group, one repeat task and one owner. Write a one-page data rule, run it for four weeks, then widen it based on weekly use and hours saved.

Is the Claude Team plan worth it for a small team?

It is worth it when two or more people use Claude on shared work or customer data. For one person, Pro is enough, and the Team plan only pays off once you need shared projects, central billing or admin controls.

Get new resources in your inbox

Consulting

Grow your business with a founder who’s done it

David coaches entrepreneurs and advises companies as a consultant and fractional Chief AI Officer.

Speaking

A speaker your audience will remember

Motivational keynotes & workshops on entrepreneurship, AI, financial literacy, and more.