Workshop · 4 or 8 hours
Online or in person, for teams of 5 to 20 people. Modules 1 to 4 in 4 hours, or all 6 modules in 8 hours.
Training · Claude for teams
Claude in your company, the right way.
The team learns by using it on real work: each person leaves with one task from their own routine running in Claude, and the company leaves with clear usage and data protection rules.
José Paulo Fernandes Júnior · Industrial engineer (UFTM), MBA from FGV · Process + AI
The program in 6 modules
Module by module1 / 6
Module 1 · Fundamentals and ground rules
- What Claude does well and where it goes wrong
- Which plan fits the company, and why
- The company’s data rule, with cases from your own area
Walks away with: The company’s AI usage policy, filled in
Module 2 · Asking well: prompts and Projects
- The request with role, context, task and format
- Refining the answer in the same chat, without starting over
- What goes in the Project instructions and what goes in the files
Walks away with: 1 Project for their own area, with instructions
Module 3 · Skills: the SOP the AI follows
- From the SOP or weekly task to a Skill
- The description that makes Claude use the Skill at the right time
- The 3-case test: 1 that passes and 2 with problems
Walks away with: 1 working Skill, built from an SOP
Module 4 · Documents, spreadsheets and reports
- Contracts and PDFs read with the clause quoted
- Spreadsheets checked line by line, with formulas
- What to check before using what the AI delivered
Walks away with: 1 real report or reconciliation redone with Claude
Module 5 · Connectors and automation
- Drive, Gmail and calendar connected safely
- Chat, Project, connector or automation: which the task needs
- Frequency, time, errors, sensitive data and owner of each task
Walks away with: A map of the 3 tasks worth automating
Module 6 · Rolling it out in the company
- Where to start, from the map of 3 tasks
- One owner per routine and a before-and-after measure
- Keeping the usage policy alive after the training
Walks away with: Rollout checklist and first-month plan
When it makes sense
- Everyone uses AI their own way, on personal accounts, with no rules at all.
- The company already pays for Claude, but hardly anyone uses it beyond asking questions.
- Customer data goes into AI without anyone knowing what is allowed and what is not.
- The same instructions get typed every time, by every person.
- The team took a recorded course, but nothing changed in the daily routine.
One team task, from request to resultProject · Finance
01Request
Skill · Payment reconciliation
Match the October payments spreadsheet against the bank statement and list only what does not match.
Each area with its own Project (a folder with the area’s instructions); the recurring routine becomes a Skill (an SOP Claude follows).
02Files
payments-october.xlsx
statement-october.pdf
Names replaced with Customer A, B and C before attaching
A company account and a one-page usage policy, agreed with the team.
Everyone knows, before pasting, whether the data can go to the AI.
03Result
Customer A R$ 1,240.00 Matches Customer B R$ 860.00 Not on the statement Customer C R$ 415.50 Paid twice Checked by the requester before sending
Claude in daily work: reports, checks and customer replies.
Practice on each area’s routine, gathered before the session.
Demo. In the training, the task comes from the team’s own routine.
Usage rules and data protection
Green · can go in
- Public text
- Blank template
- Draft with no personal names
- How-to questions
- Spreadsheet already de-identified
Yellow · only after cleanup, on the business plan
- Customer or employee data
- Contracts, proposals
- Company financial figures
- First: replace name, ID and phone with "Customer A"
Red · never goes in
- Passwords, tokens, access keys
- Full card and bank account numbers
- Sensitive personal data
- Documents under contractual confidentiality
- Anything the customer asked not to share
Company account, not personal: on business plans, Anthropic states that conversations do not train the models. And every answer that goes to a customer or a system passes through a person first.
How the training works
Training agendaClaude for teams
Before the session
- 01Gather
Before the session, 3 to 5 real tasks from each area: reports, checks, customer replies, meeting minutes.
During the session
- 02Organize
Projects per area, with the instructions and files each task needs.
- 03Practice
Spreadsheets, PDFs, documents and connectors (links to Drive and Gmail) used on the gathered tasks.
- 04Standardize
Each person turns a routine into a Skill, with its check point.
To wrap up
- 05Agree on the rules
Which account to use, who approves and how to check the answer before using it.
Assisted rollout
Within 30 days, priority routines become Skills and Projects, the usage policy is written down and the team gets follow-up. For an assisted rollout in up to 3 areas and a group of up to 20 people. Beyond that, the timeline comes in the proposal.
Technical teams
For developers: Claude Code (the Claude that works on files and spreadsheets on the computer) set up for the team, with project instructions, permissions and Skills.
Frequently asked questions
Do people need to code?
No. The workshop is for people who work with spreadsheets, documents, email and WhatsApp. Coding only comes in on the technical team track.
Which Claude plan does the company need?
It depends on team size and the kind of data. The training starts with that choice and what changes between personal accounts and business plans.
The team already uses ChatGPT. Does this still help?
Yes. The process side (written task, checks, usage rules) applies to any AI, and the training shows when each one works best.
Is it online or in person?
Both. Online anywhere; in person by arrangement, based in Batatais, Brazil.
How much does it cost?
The price depends on team size and format, and comes after a short WhatsApp conversation.
Is JP AI Systems affiliated with Anthropic?
No. This is independent training; Claude is a trademark of Anthropic.
Not affiliated with Anthropic. Claude is a trademark of Anthropic.