Interactive Demo

Configure Claude for your organization

Pick a use case, toggle features on or off, and see exactly which admin settings your deployment needs. The checklist on the right tracks what to configure — and in what order.

Preset

Chat

Enable or disable the core Claude chat experience for your organization.

Organization › Capabilities › Chat

Cowork

Computer-use agent that can edit files, and run commands on the user's machine.

Sensitive Data: MediumDownstream Actions: High
Cowork › Enable for your organization
Dispatch
Persistent agents that autonomously control the user's computer.
Cowork › Enable Dispatch
Act Without Asking
Users can let Claude act without approval — using tools, editing files, and browsing.
Cowork › Permissions › Allow 'Act without asking' mode
Always Allow Connector Tools
Let members choose "Always allow" for approving connector tools.
Cowork › Permissions › Allow 'Always allow' for connector tools
OpenTelemetry Monitoring
Send Cowork events to an OpenTelemetry endpoint for observability.
Cowork › Monitoring › OTLP endpoint
OTLP endpoint
OTLP protocol
OTLP headers
Resource attributes

Code Execution & File Creation

Allow Claude to execute code in a sandbox and create/edit documents, spreadsheets, presentations, PDFs, and data reports.

Sensitive Data: LowDownstream Actions: High
Capabilities › Code execution › Code execution and file creation
Network Egress
Give the code execution sandbox network access.
Capabilities › Code execution › Allow network egress
Domain Allowlist
Restrict which domains the sandbox can access.
Capabilities › Code execution › Domain allowlist
Base setting
Additional domains
e.g. example.com

Skills

Custom prompt-and-code bundles that extend Claude's capabilities. Skills can contain executable code and can be managed by admins or created by users.

Untrusted Input: MediumDownstream Actions: Low
Libraries › Skills › Skills
User-Created Skills
Allow team members to upload or create their own skills.
Libraries › Skills › User-created skills
Skill Sharing
Allow members to share skills with each other.
Libraries › Skills › Skill sharing
Org-Wide Skill Sharing
Allow members to share skills with the entire organization.
Libraries › Skills › Share with organization
Organization Skills
Admin-managed skills available to all org members.
Libraries › Skills › Organization skills
Add approved skill
Add

Data & Privacy Controls

Controls for data sharing, feedback, and visibility within the organization.

Data and privacy
Rate Chats (Feedback)
Allow people to rate Claude's responses and share that feedback with Anthropic.
Data and privacy › Rate chats
Share Chats
Allow people to share chats with others in your org.
Data and privacy › Share chats
Share Connector Chats
Allow people to share chats that use connectors. Recipients see Claude's response but not the connector data.
Data and privacy › Share chats that use connectors
Location Metadata
Allow Claude to use coarse location metadata (city/region) to improve product experiences.
Data and privacy › Location metadata
Public Projects
All users in an organization can see and start chats in public projects.
Data and privacy › Public projects
Data Retention
Configure how long chats and uploads are retained.
Data and privacy › Retention period
Separate Retention Periods
Set different retention periods for chats vs projects.
Data and privacy › Separate retention periods
Chat retention period
Project retention period
DATA SOURCES
VISUALS
INTEGRATIONS
ADMINISTRATION

Claude use cases by industry — and the tenant configurations that matter

We've mapped the most common Claude use cases across six industries — and exactly which admin settings each one requires, which configs break functionality if disabled, and where the risks sit.

Configs reassessed on every platform updateCovering Claude Chat, Cowork & CodeRisk breakdowns for connectors, skills, plugins & custom combinations
Specific use cases in mind? Get exact tenant configs — required enables, settings that break functionality, and risks.

Use cases by industry

Select your industry. These are the use cases we see most — each carries a different configuration profile.

Contract Redlining & Review
Automated clause analysis against negotiation playbooks, deviation flagging, and redline generation for vendor contracts, NDAs, and customer agreements.
Legal Research & Memo Drafting
Case law analysis, regulatory research, and structured memo generation. Requires careful attention to data residency and confidentiality settings.
Due Diligence Acceleration
M&A document review, risk identification across large document sets, and structured reporting for deal teams.
NDA Triage & Classification
Rapid screening of incoming NDAs — standard approval, counsel review, or full legal review — routed to the appropriate tier automatically.

What you get: configuration breakdown by use case

For each use case you select, we map every relevant admin setting to a clear recommendation. Here's a preview.

Use CaseRequired EnabledCannot Be DisabledRisks
Contract Redlining
Cowork
Organization > Libraries > Skills > Cloud Code Execution
Organization > Libraries > Skills > Skills
Organization > Libraries > Skills > Cloud Code Execution
Cowork > New Chat > Work in a folder
Confidential clause exfiltration via injected skill; prompt injection via untrusted document content
Legal Research
Cowork, Claude in Chrome
Organization > Claude in Chrome > Enable for your Team
Organization > Capabilities > Data Sources > Web Search
Organization > Capabilities > Data Sources > Web Search
Organization > Claude in Chrome > Enable for your Team
Prompt injection via untrusted legal databases; session hijacking on privileged sites if blocklist not configured
Financial Report Analysis
Chat, Cowork
Organization > Capabilities > Code Execution > Cloud Code Execution
Organization > Capabilities > Code Execution > Allow Network Egress > Package Managers Only
Organization > Capabilities > Code Execution > Cloud Code Execution
Organization > Libraries > Skills > Skills
MNPI exposure if data residency misconfigured; network egress enables exfiltration of financial data
Clinical Documentation
Cowork
Organization > Cowork > Monitoring
Organization > Libraries > Skills > Cloud Code Execution
Organization > Cowork > Enable for your Organization
Cowork > New Chat > Work in a folder
PHI exfiltration via prompt injection; dispatch agents operating unattended on systems with patient data
Code Review
Cowork, Claude Code
Organization > Capabilities > Code Execution > Cloud Code Execution
Organization > Capabilities > Code Execution > Allow Network Egress > Package Managers Only
Organization > Capabilities > Code Execution > Cloud Code Execution
Cowork > New Chat > Work in a folder
Source code exfiltration via curl in sandbox; supply chain injection via unvetted marketplace plugins
Supply Chain Analysis
Cowork
Organization > Capabilities > Code Execution > Allow Network Egress > Package Managers Only
Organization > Libraries > Skills > Skills
Organization > Libraries > Skills > Cloud Code Execution
Organization > Capabilities > Code Execution > Allow Network Egress
Proprietary supplier data leakage via unrestricted network egress; competitive intelligence exposure
Student Data Analysis
Cowork, Chat
Organization > Capabilities > Data Sources > Ask Organization
Organization > Capabilities > Code Execution > Cloud Code Execution
Organization > Capabilities > Data Sources > Ask Organization
Organization > Capabilities > Code Execution > Cloud Code Execution
Student PII aggregation if Ask Organization ingests unclassified data; FERPA violation via shared chats

Why configuration matters

Every Claude use case carries a different configuration profile. Settings interact in non-obvious ways, and the wrong combination can either break functionality or expose your organization to risk.

Settings have hidden dependencies

Skills require Organization > Libraries > Skills > Cloud Code Execution to be enabled — without it, skills and plugins won't run. But enabling cloud code execution also opens network egress by default, which means the sandbox can reach external domains unless you explicitly restrict it. A team deploying contract redlining skills without configuring the egress allowlist has an open exfiltration path they may not know about.

The same setting means different things for different use cases

Organization > Claude in Chrome > Enable for your Team is essential for legal research workflows that need to navigate case law databases. But for clinical documentation, where patient portals are in the browser, enabling Claude in Chrome without a strict blocklist means Claude can access those sessions with the user's credentials. Same toggle, completely different risk profile.

Risk compounds when you run multiple use cases

A tenant running code review alongside financial analysis needs network egress for package installs, but that same egress path is a risk for the financial analysis use case handling MNPI. Multi-use-case tenants need configurations scoped to the most sensitive workflow, not the most convenient one.

Stop guessing at configurations

Tell us your use cases. We map every relevant setting path to a clear recommendation — what to enable, what breaks functionality if disabled, and the specific risks for your combination.