Claude in the Company – Overview
Claude is introduced here not as a single tool, but as a controlled company platform: a secure base, structured team work, and carefully reviewed extensions. This section explains how we use it safely and get real value from it.
The four-layer operating model
Layer 1 Governance & secure base → Enterprise: SSO, SCIM, roles, audit logs, retention
Layer 2 Structured team work → Projects, Skills, Artifacts
Layer 3 Controlled extension → Connectors, Plugins (only after review)
Layer 4 Technical integration → Platform / API, automation
Recommended sequence: Enterprise first → Teams & Projects → Skills → Connectors → Cowork → Claude Code broadly only with technical governance.
The golden rules
The whole framework reduces to a few habits everyone shares:
- Only approved data. Mind the 🟢/🟡/🔴 data traffic light — never 🔴 patient, health, or credential data.
- Only the company access. No private accounts for work content.
- Always check results. Claude drafts; a human decides and signs off.
- Only approved extensions. Connectors, plugins, and MCPs only after review.
At a glance
Where to go next
| If you are… | Start with | Why |
|---|---|---|
| Any employee | Onboarding – First Steps | Get productive and safe in one sitting |
| A team lead / owner | Adoption & Governance Guide | Set up Projects, Skills, and team governance |
| An administrator | Training Plan → Schulung 4 | Configure Enterprise: SSO, SCIM, go-live |
| Planning trainings | Training Plan | Four trainings, trainer guidance, rollout schedule |
Also in this section
- Surfaces & Features Guide — where Chat, Projects, and Claude Code differ
- Artifacts Guide — reusable deliverables in Claude Chat
Keep learning the craft
For practical, bite-sized lessons on prompting, tone, and getting more from Cowork, Skills, and Connectors, claude101.com is a useful external complement. The internal rules and guides here always take precedence.
All AI-generated content is a draft and requires human review before external use.