Claude for business: Team and Enterprise plans
Claude for business means one of two Anthropic plans: Team, billed per seat, or Enterprise, billed as a seat fee plus usage. Both give your company control of every account, and neither uses your conversations to train models by default. If a few of your people already pay for Claude on their own cards, Team brings them and their work under one company account. This guide explains what each plan includes, how each charges and which one fits a manufacturer or distributor. It then covers the first jobs to give Claude, the settings to change on day one and how to roll it out to your team.
Claude plans at a glance
Anthropic sells Claude as individual plans for one person and company plans for a team. The table shows who each plan is for and how it charges, from Anthropic’s pricing page and help centre on October 5, 2026. Current rates are on that page, and Anthropic’s Team plan article says pricing and currency vary by region.
| Plan | Who it is for | How it charges | Company controls |
|---|---|---|---|
| Free, Pro and Max | One person on a personal account | Free has no fee. Pro is billed monthly or yearly, and Max monthly. | None. Each person decides whether their chats can be used to train models. |
| Team, Standard seat | Companies with 2 to 150 people on Claude | A fee per seat each month, billed monthly or yearly | An admin console, single sign-on and connector controls. No model training on your content by default. |
| Team, Premium seat | Heavy users, such as an estimator who works in Claude most of the day | A higher fee per seat, for five times the usage of a Standard seat | The same as a Standard seat |
| Enterprise | Companies that need audit logs, automatic account provisioning or custom retention | A seat fee for access, plus all usage at API rates, billed yearly | Everything in Team, plus audit logs, SCIM, custom data retention and role-based access |
On the individual plans, each person controls a privacy setting that decides whether Anthropic may use their chats to improve its models. Anthropic’s pricing page lists model training on those plans as opt-out. With the setting on, Anthropic may keep those chats in de-identified form for up to five years, according to its consumer retention article. Company records belong on Team or Enterprise.
The Claude Team plan
Team needs at least two members and takes up to 150 seats, according to the Team plan article. The person who creates it must use a business email address, because public domains such as Gmail are not accepted for that account (setup guide).
Anthropic’s pricing page lists what each seat includes. Each person gets Claude on the web, the desktop and phones, plus Claude Code, Anthropic’s coding tool. Claude also works in Excel, Word, PowerPoint and Outlook. Owners get central billing, single sign-on, control over which connectors staff can use, and usage analytics. Anthropic does not train its models on Team content by default.
Team charges a fee per seat each month. You choose monthly or yearly billing, and Anthropic’s pricing page says yearly plans are designed to cost less. A Standard seat gives more usage than an individual Pro plan, and a Premium seat gives five times the usage of a Standard seat. You can mix the two, so Premium seats go only to the people who need them.
Usage limits apply to each person, so one person reaching a limit does not stop anyone else (Team plan article). Limits run on a rolling five-hour window, with a weekly limit on top, according to the pricing page. An owner can buy usage credits so people keep working past their limit, at API rates. The owner can also set a monthly spend limit for the company and for each person.
Claude Enterprise
Enterprise includes everything in Team and adds security and compliance controls. Anthropic’s Enterprise plan article lists these additions:
- Audit logs of user actions, system events and data access.
- SCIM (System for Cross-domain Identity Management), which creates and removes Claude accounts from your identity provider.
- Custom data retention, which deletes chats and projects after a period you set, with a minimum of 30 days.
- Compliance and analytics APIs that let your own systems pull activity logs and usage figures.
- US-only inference, which keeps the model’s processing in the United States.
- Customer-managed encryption keys, held in your own cloud account.
Enterprise also charges differently. The seat fee covers access only. All usage in Claude and Claude Code is billed separately at API rates, with no included allowance and no limit per seat. Admins set spend limits for the company and for each person, and the plan is billed yearly. API rates are what Anthropic charges developers for the text each model reads and writes, and Claude API pricing lists them by model.
You can buy Enterprise online with at least 20 seats, paying in US dollars and buying usage credits up front. Through Anthropic’s sales team the minimum is 50 seats, with invoicing, other currencies and usage billed monthly after it is used.
Which plan fits your business
For a manufacturer or a distributor, four questions settle it.
- How many people need Claude? Team covers 2 to 150 seats. Above 150, Anthropic points Team customers to Enterprise.
- Does a customer, lender or auditor ask for audit logs or automatic account removal? Team has neither, according to the plan comparison, so you need Enterprise.
- Do a few people use Claude all day? On Team, give them Premium seats and keep everyone else on Standard. On Enterprise, each person’s usage is billed as it happens.
- Do some records have to stay in a particular country? Anthropic stores data in the US on both plans. On Enterprise, the one location setting keeps processing in the United States.
If none of those points to Enterprise, start on Team. Anthropic supports moving an organization from Team to Enterprise later.
Get your team using the seats you pay for
Tell Derik how many people use Claude today and which job you want it to take on first. He will tell you which plan and settings fit.
Start a conversationFirst jobs to give Claude
Start with jobs built on documents your team already handles, where a person checks the result before it goes anywhere. These six suit a manufacturer or a wholesale distributor.
| Job | What Claude does | What a person checks |
|---|---|---|
| Replies to quote requests | At a machine shop, drafts a reply in Outlook from the customer’s email and your price list. The draft waits until someone sends it. | Price, lead time and terms against the ERP |
| Supplier quotes side by side | At an industrial distributor, reads four supplier quotes and builds one table of price, lead time, freight and exclusions | Each total against the original PDF |
| ERP exports in Excel | At a building-products distributor, answers questions about a sales export, such as which customers bought less than last year, and writes formulas you can follow | The total against a report you already trust |
| Customer purchase order terms | At a food manufacturer, lists where a grocery customer’s purchase order terms differ from your terms of sale, such as payment dates or deductions | Each difference, by whoever signs contracts |
| Work instructions | At a beverage plant, turns a supervisor’s notes on a line changeover into a written procedure, in English and French | The steps, by the supervisor who runs the line |
| Product questions from sales | At a distributor, answers from a shared project that holds the catalogue, spec sheets and pricing rules | That the project’s files are current, by a named owner |
Anthropic’s Microsoft 365 page describes the Outlook and Excel features used here, and lists Claude for Outlook as a beta. Leave out, for the first month, anything that sends, orders or changes ERP records without a person approving it. Human in the loop covers how to set up that approval when you get there. When a question needs the ERP itself rather than an export, see AI reporting.
Settings to change on day one
A Team plan works with Anthropic’s defaults on the day you buy it. Change these settings before you invite the team. Most sit under Organization settings, and an Owner or the Primary Owner can change them.
- Verify your domain and restrict organization creation. Once your email domain is verified, the Restrict organization creation setting stops staff from opening new Claude accounts on it, including personal ones (SSO guide).
- Set up single sign-on and require it. With it required, staff sign in through your identity provider, such as Microsoft Entra ID, Google or Okta.
- Decide on feedback. The thumbs up and down buttons send the whole conversation to Anthropic. Anthropic keeps it for up to five years and may use it to train models. Owners can turn this off with the Rate chats setting.
- Turn on only the connectors you need. An owner enables each connector for the company, and can let it read from a service without writing back to it. Claude reaches only what each person can already see in that service (connectors guide).
- Decide how projects are shared. Project sharing and public projects are both on by default. Turn public projects off if every project should be shared with named people.
- Write organization instructions. Up to 3,000 characters of instructions apply to every conversation. Use them for Canadian spelling, metric units and a reminder to flag assumptions.
- Set spend limits with usage credits. When you turn credits on, set a monthly limit for the company and for each person.
- Keep company work in shared projects. When someone leaves, the rest of the team loses access to their chats and private projects. The Primary Owner can still export them (offboarding guide).
None of these settings changes where your data sits. Anthropic’s server location article says data is stored in the US. By default, requests may be routed to the US, Europe, Asia and Australia. On usage-based Enterprise plans, a US-only inference setting keeps processing in the United States, at a higher rate for newer models. If some records have to stay in Canada, private AI for business covers the options. For the rest of your setup, secure AI at work has a checklist.
How to roll Claude out to your team
Roll Claude out in four steps, in this order.
Write a one-page AI policy
Before anyone signs in, tell staff which account to use and which records may go into Claude. Name who checks the output before it reaches a customer or a supplier. One page is enough to start, and AI policy has a template written for manufacturers and distributors.
Bring personal accounts into the company account
Staff with a personal Claude account on their work email can move it into your Team workspace and bring their chats, projects and files. Custom skills do not move, and connected apps have to be reconnected. A Pro or Max plan bought directly is cancelled with a prorated refund, according to Anthropic’s migration guide. On Team, each person starts their own move, and on Enterprise, admins can also claim existing accounts on the company domain.
Train each desk on its own work
Train each desk on the documents it handles every day, such as the estimator on quote requests and the buyer on supplier quotes. AI training covers how to run those sessions, and the Ontario and Quebec programs that can pay for employer training.
Run one first project and measure it
Pick one job from the table above, name an owner and record how long it takes today. Measure it again after two weeks. Check the usage analytics in the admin console to see who uses Claude. Move Premium seats to the people who use them most, then pick the next job.
Claude versus ChatGPT for a business
On data handling, the basics match. Anthropic does not train its models on business data by default, according to its training article. OpenAI says the same of ChatGPT Business, Enterprise and Edu workspaces in its security documentation. Both remove a deleted chat from their systems within 30 days, with exceptions for legal and safety reasons. Anthropic says so in its retention article, and OpenAI in its ChatGPT Work overview.
The differences that matter to a manufacturer or distributor are these:
- Office tools. Claude works inside Excel, Word, PowerPoint and Outlook, with edits shown as tracked changes or highlighted cells, according to Anthropic’s Microsoft 365 page.
- How the large plan charges. Claude Enterprise adds usage at API rates to a seat fee. OpenAI’s admin guide describes ChatGPT Enterprise agreements where eligible usage draws on a shared pool of credits or is billed in US dollars.
- Where data sits. Anthropic stores data in the US. OpenAI’s API data guide lists a Canada region that stores API data at rest and does not process it there. ChatGPT alternatives for business compares storage plan by plan.
To choose, give the same three jobs to a few people on each tool for two weeks and compare the drafts. Then put the company on one account. Two company AI accounts mean two sets of settings to keep and two accounts to close whenever someone leaves.
How ThriveAI helps
ThriveAI is an AI engineering company in Ottawa. It builds private AI systems on a company’s own data, for businesses that make, move or sell physical goods. Its Claude and ChatGPT rollout covers which seats to buy, which jobs to start with and which settings to change. ThriveAI also runs AI training on site, built on your own documents and systems. It writes an AI policy your team can follow, and sets the account settings that enforce it.
For records that have to stay in house, ThriveAI’s platform is designed to keep each client’s data on its own server in Canada. You choose a model on that server or a hosted model under a written zero data retention agreement. A hosted model may process requests outside Canada. Derik Lawlis, the founder, leads every project and stays close to the build.