Documentation

Organization Policies - Detailed Guide

View as Markdown

Organization policies provide sophisticated control over member access to features and settings. Understanding how policies work, their enforcement modes, and how they affect individual settings is essential for effective organization management.

Policy Enforcement Modes

Organizations can choose between two enforcement modes that determine how policies affect member settings.

Strict Enforcement Mode

Strict Mode (enforceStrict: true): Policies are mandatory and cannot be overridden

Characteristics:

  • Mandatory: Organization policies override all member settings
  • No Override: Members cannot change settings that conflict with policies
  • Complete Control: Administrators have full control over all settings
  • Compliance: Perfect for highly regulated environments

When to Use:

  • Regulatory compliance requirements
  • Security-critical environments
  • Organizations requiring strict standardization
  • Audit and compliance scenarios

Example: If organization policy sets enhancedSearchEnabled: false in strict mode, members cannot enable enhanced search, regardless of their account settings.

Non-Strict Enforcement Mode

Non-Strict Mode (enforceStrict: false): Policies provide defaults that members can modify

Characteristics:

  • Defaults: Policies provide default settings
  • Member Control: Members can modify settings (with restrictions)
  • Flexibility: Balances control with user autonomy
  • Less Permissive Rule: Members can only make settings less permissive

When to Use:

  • Organizations wanting guidance without strict control
  • Teams needing flexibility while maintaining standards
  • Gradual policy implementation
  • Collaborative environments

The Less Permissive Principle

In non-strict mode, the system follows a "less permissive" rule that determines what changes members can make.

Understanding Permissiveness

Permissive Settings: Settings where true = enabled/allowed and false = disabled/restricted

Examples of Permissive Settings:

  • chatEnabled: true = AI Chat enabled (more permissive), false = disabled (less permissive)
  • enhancedSearchEnabled: true = enhanced search enabled (more permissive), false = disabled (less permissive)
  • ocrEnabled: true = OCR enabled (more permissive), false = disabled (less permissive)

The Rule

Members Can: Change true → false (disable features that organization allows)

Members Cannot: Change false → true (enable features that organization restricts)

Examples

Example 1: Organization Allows, Member Disables

  • Organization Policy: enhancedSearchEnabled: true
  • Member Setting: Can set to false (disable for their account)
  • Result: ✅ Allowed - Member is making it less permissive

Example 2: Organization Restricts, Member Tries to Enable

  • Organization Policy: enhancedSearchEnabled: false
  • Member Setting: Cannot set to true (enable for their account)
  • Result: ❌ Not Allowed - Member cannot make it more permissive

Example 3: Strict Mode

  • Organization Policy: enhancedSearchEnabled: true, Strict Mode: true
  • Member Setting: Cannot change (locked to organization policy)
  • Result: ❌ Not Allowed - Strict mode prevents all changes

Policy Indicators

Visual indicators throughout the application show when organization policies affect settings.

Policy Indicator Icons

Policy Indicator: Small icon next to settings indicates organization policy control

Indicator States:

  • Present: Policy affects this setting
  • Tooltip: Hover for details about the policy
  • Color Coding: Different colors indicate policy type or enforcement level

Understanding Indicators

Organization Controlled: Setting is controlled by organization policy

Organization Default: Setting uses organization default but can be modified (non-strict)

Strict Enforcement: Setting cannot be changed (strict mode)

No Policy: No organization policy affects this setting

How Policies Affect User Settings

Policy Application Flow

When a user views or modifies settings:

  1. Load User Setting: System loads user's current setting
  2. Check Organization Policy: System checks if organization has a policy for this setting
  3. Apply Policy: Policy is applied based on enforcement mode
  4. Display Effective Setting: User sees the effective setting (policy-applied value)
  5. Show Indicators: Policy indicators show policy influence

Strict Mode Behavior

In strict mode:

  • Override: Organization policy value replaces user setting
  • Locked: User cannot change the setting
  • Visual: Settings show as disabled/locked with policy indicators
  • Immediate: Changes take effect immediately for all members

Non-Strict Mode Behavior

In non-strict mode:

  • Default: Organization policy provides default value
  • Modifiable: User can change setting (if making it less permissive)
  • Visual: Settings show policy indicators but remain editable
  • Flexible: Members have control within policy constraints

Policy Categories

Governance Policy

Governance extends organization policy into AI execution retention, replay, and evidence without changing how members normally use chats or workflows.

Settings:

  • preventChatDeletionWhenGoverned: Blocks deletion of AI chats when organization policy requires retention
  • preventWorkflowDeletionWhenGoverned: Blocks deletion of AI workflows and AI workflow jobs when organization policy requires retention
  • archiveContentInsteadOfDelete: Enforces content archiving so delete actions remove content from normal view while retaining source artifacts for governance references

How It Works:

  • Clear Ideas writes governed AI artifacts to secure storage for AI chats, AI workflows, AI workflow jobs, and governed MCP evidence
  • MongoDB remains the hot operational store for active usage and can be slimmed later without losing governed evidence
  • Organization policy determines deletion-prevention behavior and is stamped into governed AI records
  • Public chat uses the same storage architecture, but organization policy primarily applies to organization-owned AI activity
  • Content archiving helps governed evidence keep resolving source artifacts even when content is removed from ordinary site views

Policy Versioning and Hashes

Organization policy now includes:

  • version
  • policyHash
  • publishedAt

Every governed AI record stores the policy version and hash that were effective when the record was materialized. This makes it possible to answer which policy governed a chat interaction or workflow run at a specific point in time.

Policy History

Every policy update creates a versioned policy-history record. Organization administrators can review older policy versions and use them as supporting evidence when analyzing governed AI activity.

Governed AI Storage Model

Governed AI uses a write-through model:

  1. A chat interaction or workflow result is assembled in memory
  2. The exact bytes that will be stored are serialized once
  3. A hash is computed over those exact bytes
  4. The artifact is written to governed storage
  5. The application record keeps lightweight metadata such as sequence, hash, and storage pointer

Governed chat storage is event-based, so later interactions append compact records rather than rewriting the full chat every time. This avoids canonical JSON issues and supports later replay, export, and forensic review.

Prompt Versioning

AI Chat interactions also stamp the prompt context used for that interaction, including:

  • system prompt key
  • system prompt version
  • system prompt hash
  • effective prompt hash

This allows a continued older chat to use the latest active Clear Ideas prompt while still preserving evidence of which prompt governed each later interaction.

Evidence and Export

Governed evidence is primarily for audit, replay, and forensic analysis.

  • Normal application usage continues to rely on the live chat and workflow records
  • Governed artifacts remain available for controlled export when needed
  • Organization administrators retrieve governed evidence from Audit & Governance > Governance Activity
  • Evidence export is intended for controlled review, not day-to-day browsing
  • Customer-visible evidence exports include prompt references and hashes rather than raw proprietary Clear Ideas prompt bodies
  • Signed evidence exports include Ed25519 receipts, file hashes, and signing key metadata when evidence signing is enabled
  • Administrators can verify a downloaded evidence ZIP locally in the browser with Verify Evidence Export

Signed export verification helps reviewers confirm that the exported evidence files still match the signed receipt after the ZIP leaves Clear Ideas. The verifier runs client-side, so the evidence bundle is not uploaded during the check.

Governance Review and AI Usage

Organization administrators now have two complementary governance views:

  • Governance Activity: Review governed AI lifecycle activity, export evidence for AI Chat, AI Workflow, AI Workflow Job, and governed MCP records, and verify downloaded signed evidence bundles from one centralized table
  • AI Usage: Analyze AI credits, tokens, model usage, and usage by organization member across AI Chat and user-owned workflows

These views are intentionally separated from everyday object menus so that audit, review, and evidence retrieval stay centralized.

AI Features Policy

Control AI capabilities across the organization:

Settings:

  • clientEnabled: Enable/disable AI Chat (master switch)
  • chatEnabled: Enable AI Chat for sites
  • summariesEnabled: Enable AI document summaries
  • enhancedSearchEnabled: Enable AI enhanced search
  • mcpEnabled: Enable External Tool Usage (MCP)
  • useCreditsForThirdParty: Allow organization billing for collaborator AI usage
  • permittedModels: Specify allowed AI models (empty = all allowed)

Policy Indicators: Appear on AI settings pages showing organization control

Model Policy

Permitted model policy supports:

  • All models: leave permittedModels empty
  • Latest aliases: allow Clear Ideas-managed workflow model aliases such as the current intelligent or latest production options
  • Specific models: allow named model versions for stricter repeatability
  • Deprecated models already in policy: show only so administrators can unselect them

Policy is static. Clear Ideas does not automatically select a successor when a deprecated model is no longer permitted or no longer available. AI Chat hides unavailable choices. AI Workflows show a discreet validation warning when a workflow or step references a deprecated model, and workflow jobs fail with an explanatory message if execution reaches a model that policy does not permit.

Content Deletion Policy

When content archiving is enabled by organization policy, site delete actions can be converted into archiving instead of permanent deletion. Archived content is hidden from normal site browsing and can be restored by administrators, while the underlying file and content references remain available for governance purposes.

At the site level, content deletion behavior has three modes:

  • Allow Content Deletion: delete actions use Recently Deleted where deletion is allowed
  • Archive Content: delete actions move content into Archived Content
  • Block Content Deletion: delete and archive actions are hidden and blocked

In non-strict mode, sites can choose a less permissive mode. For example, if organization policy enforces archiving, a site may block content deletion entirely, but it cannot allow ordinary deletion. In strict mode, the organization policy setting is locked.

Search Policy

Control search capabilities:

Settings:

  • fullTextSearchEnabled: Enable full-text search
  • ocrEnabled: Enable OCR for PDF documents

Policy Indicators: Appear on search settings pages

Signing Request Policy

Control signature request behavior and signing terms:

Settings:

  • requestsEnabled: Allow or block members from creating, editing draft, and sending signature requests
  • defaultDisclosureBody: Organization-level markdown signing terms applied to member signature requests
  • allowUserDefaultDisclosureOverride: Allow members to replace the organization default with their own signing request default

How It Works:

  • If requestsEnabled is false, members cannot create or send signature requests
  • If defaultDisclosureBody is set and overrides are disabled, the organization terms are enforced for all member requests
  • If defaultDisclosureBody is set and overrides are allowed, members may use their own default signing terms instead

Policy Indicators: Appear in signing request defaults and related signing UI when organization policy affects those settings

Notification Policy (Strict Mode Only)

Control notification settings (only applies in strict enforcement mode):

Settings:

  • frequency: Notification frequency (daily, weekly, monthly)
  • hours: Notification hours (array of hours 0-23)
  • days: Notification days (array of day names)

When Active: Only enforced when enforceStrict: true

Site Policy (Strict Mode Only)

Control site-level settings (only applies in strict enforcement mode):

Settings:

  • autoAcceptInvites: Automatically accept site invitations
  • enableLocalSync: Enable local synchronization

When Active: Only enforced when enforceStrict: true

Mandatory AI Instructions

Organization-level AI instructions that are automatically included with all member AI interactions.

Characteristics

Mandatory: Applied automatically to all organization members

Cannot Be Disabled: Members cannot disable organization instructions

Combined with Personal: Organization instructions are combined with personal instructions

Limit: Up to 10 mandatory instructions per organization

Policy Indicators

Mandatory instructions are indicated in AI Chat interfaces, showing that organization instructions are active.

Viewing Policies as a Member

Non-admin members can view organization policies:

Policy Summary

Navigate to Settings > Organization > Organization Policies to view:

  • Policy Summary: Overview of active policies
  • Policy Indicators: Visual indicators showing policy control
  • Effective Settings: How policies affect your settings
  • Enforcement Mode: Whether strict or non-strict mode is active

Understanding Your Settings

Policy Indicators: Show which settings are controlled by organization

Effective Values: See the actual settings that apply (after policy application)

Change Limitations: Understand what you can and cannot change

Policy Update Process

Updating Policies

Organization administrators update policies:

  1. Navigate to Settings > Organization > Organization Policies
  2. Modify policy settings
  3. Click Save Policy
  4. Changes take effect immediately

Impact of Policy Updates

Immediate Effect: Policy changes apply to all members immediately

Member Impact: Members may see settings change automatically

Notification: Consider notifying members of significant policy changes

Last Updated: Policy update timestamp is tracked for audit purposes

Policy Inheritance and Precedence

Hierarchy

Settings are determined by this hierarchy (highest to lowest):

  1. Organization Policy (Strict Mode): Overrides everything
  2. Organization Policy (Non-Strict Mode): Provides defaults with restrictions
  3. User Account Settings: User's personal preferences
  4. Site Settings: Site-specific overrides (subject to account and organization policies)

Precedence Rules

Strict Mode: Organization policy always wins

Non-Strict Mode:

  • Organization provides default
  • User can make less permissive changes
  • Site settings can further restrict (but not enable if disabled higher up)

Best Practices

Policy Planning

Start Non-Strict: Begin with non-strict mode to allow flexibility while establishing defaults

Gradual Strict: Move to strict mode after policies are tested and refined

Document Policies: Document why policies are set and what they achieve

Communication

Inform Members: Communicate policy changes to members

Explain Rationale: Help members understand why policies exist

Provide Guidance: Offer guidance on working within policy constraints

Regular Review

Review Policies: Periodically review policies for relevance

Update as Needed: Update policies as organizational needs change

Monitor Compliance: Monitor how policies affect member workflows