How Do I Get a Non-Technical Team Actually Using Claude in the First 30 Days?
A week-by-week plan for getting a non-technical team using Claude chat and Cowork in 30 days: admin settings, one shared project, owner-provisioned skills, scheduled tasks and the analytics to check.
You get a non-technical team using Claude in 30 days by shrinking the job: one shared project per team, one real recurring task per person, and a weekly check of the usage numbers your admin can already see. Buying seats and sending a "you now have Claude" email does not do it. What works is replacing one piece of work each person already does, every week, inside Claude, until opening it is a habit.
Below is the plan I'd run, week by week, with the settings to check before anyone logs in.
Why seats alone don't create adoption
On team calls the opening problem is almost always the same: the company pays for seats, a handful of people use them daily, and everyone else tried it once, got a generic answer, and went back to their old way of working. That isn't a motivation problem. A blank chat box asks people to invent a use case on the spot, and most people, reasonably, don't.
So the first 30 days are not about teaching "AI." They are about giving each person a specific, boring, recurring task that Claude does well, set up so the context is already there when they open it.
Before day one: the admin settings to check
Spend an hour in Organization settings before inviting anyone. On Team and Enterprise plans, the Cowork admin guide lists the controls that matter most for a cautious rollout:
- Cowork is on by default, and owners can turn it off for the whole organization. Enterprise owners can instead enable it for specific groups through custom roles.
- "Allow 'Always allow' for connector tools" is off by default. Leave it off for the first month, so anything that writes to another system asks for approval each time.
- Auto approval mode can be allowed or blocked by an owner. For a non-technical team I'd start with manual approval, so people see each step Claude proposes before it happens.
- Web search for Cowork and Chat can be turned off under Capabilities, if your policy requires it.
Then decide which connectors you'll allow. If your policy is read-only to start (email and calendar readable, nothing sent on someone's behalf), say so in one sentence and put it at the top of the shared project's instructions. People adopt faster when they know where the line is.
Week 1: one shared project per team, and one task per person
Create one project for each team (operations, marketing, sales, finance) and share it. Projects on Team and Enterprise have two permission levels: Can view lets people chat inside the project and see its knowledge and instructions, and Can edit lets them change the instructions and knowledge. Give most of the team Can view and one or two owners Can edit, so the project stays clean.
Put three things in each project:
- Instructions that describe the team, its customers and its tone in a short paragraph or two.
- Knowledge: the handful of documents people keep re-explaining (the pricing sheet, the style guide, the process doc, last quarter's report).
- The line: what not to paste in and what stays human.
Here's the part people skip: context does not carry between chats in a project unless it's in the project knowledge. If someone explains something useful in a chat, the project owner copies it into the knowledge base. That's how the project gets smarter over the month instead of starting from zero every conversation.
Then the important part: on day one, every person names one task they did last week that they will do again this week. A status update, a meeting summary, a first draft of a customer reply, a weekly numbers recap. They do that task in the project, in chat, and compare it to how they'd normally do it. One task. Not a tour of features.
Week 2: let Claude do the multi-step version in Cowork
Once people trust the answers, move the same tasks from "help me write this" to "go do this." Cowork is the agentic mode: you describe the outcome and Claude plans and carries out the steps, with files, connectors and the browser. Chat and Cowork share one message box now, and a project's files, instructions and memory serve both (more on that in the post on the merge).
The jump I look for in week two: someone stops pasting a spreadsheet into chat and instead asks Cowork to read the file, pull the three numbers that changed, and draft the update. On the desktop app Claude can read and write files in folders you've connected, and only those, which is a good thing to say out loud to a nervous team.
Run one live session this week, as a group, where each person shows their one task. Seeing a colleague's real example does more for adoption than any demo I can give.
Week 3: turn the good prompts into shared skills
By week three a few people will have prompts that work every time: the way they ask for a customer reply, the format for the weekly recap. Turn those into skills. Skills are folders of instructions and resources that Claude loads when a task calls for them, and on Team and Enterprise plans, organization owners can provision a skill for all users so it appears automatically in everyone's skills list.
Start with two or three, not twenty. A skill called "weekly team update in our format" that everyone uses beats a library nobody opens. (Skills require code execution to be turned on, so check that setting first.)
Week 4: schedule the recurring work
The last step is the one that makes Claude part of the week instead of a tool people remember to open. Scheduled tasks run on a cadence (hourly, daily, weekdays or weekly) and run remotely, so they keep going when someone's laptop is asleep. Tasks that need local files or apps only run on that computer. You create one from Scheduled in the sidebar with New task, or just describe the task and how often it should run in a conversation and confirm what Claude proposes.
Good first candidates for a non-technical team:
- A Monday morning summary of last week's notes in the team's shared project.
- A Friday draft of the weekly status update, waiting for a human to edit and send.
- A daily digest of anything flagged in a connected tool your policy allows Claude to read.
Notice that each one ends with a person reviewing it. In the first month, scheduled work should produce drafts, not send things.
How to tell if it's working
Team plan owners and primary owners can see usage analytics in the admin console. The two numbers I'd watch weekly:
- Percentage of users with one or more chat (Analytics > Claude Chat), plus projects created and used.
- Percentage of users with one or more Cowork session, and daily, weekly and monthly active Cowork users (Analytics > Cowork).
Write both numbers down on day one. If the share of people using Claude each week isn't clearly higher by day 30, the fix is almost never more features. It's usually that some people never got a task that was really theirs. Go back to week one with them.
A one-page checklist
- Admin: Cowork on, "Always allow" for connector tools off, manual approval, connectors decided, policy line written down.
- One shared project per team, instructions + knowledge + the line, most people on Can view.
- Every person names one recurring task on day one.
- Week 2: move that task to Cowork, one group show-and-tell.
- Week 3: two or three owner-provisioned skills from prompts that already work.
- Week 4: one scheduled task per team that produces a draft for a human.
- Analytics baseline on day one, check on day 30.
If you're weighing whether a newer model changes any of this, my take on Fable 5.2 for teams covers which seats get it and what it costs. And if part of your team will work from their phones, Cowork on mobile is worth a read.
Sources: Get started with Claude Cowork, Use Claude Cowork on Team and Enterprise plans, How can I share a project, What are skills, Schedule recurring tasks in Claude Cowork, View usage analytics for Team and Enterprise plans.
If you'd rather have someone run the first session with your team, built around the tasks your people actually do, that's what our private team training is for.
Related posts
We Pay for Claude Seats and Only a Quarter of the Team Uses Them. How Do We Fix Adoption?
A recovery plan for a stalled Claude rollout: read the usage analytics you already have, sort people into three groups, give each a fix, and right-size seats instead of paying for empty ones.
Claude Cowork Got Its Own Browser. The Isolation Model Is the Real Story
Anthropic shipped a built-in browser for Claude Cowork on August 26, 2026: no extension, no shared logins, one permission card per site. Here's what actually changed and why the access model matters more than the browsing itself.
Claude Merged Chat and Cowork Into One Project. The Mode Picker Is Gone
Anthropic quietly merged Chat and Cowork into a single Home on claude.ai web and desktop in early August 2026: one composer, and a project's memory, instructions, files, and scheduled tasks are now shared across both. Here's what changed and why the removed decision matters more than any new button.