Sypha AI Docs
Plans & OrganizationEnterprise Features

Model Governance & Access Control

Strategically manage which AI engines and providers are authorised for use across your engineering department.

Model Governance & Access Control

The Model Governance module provides administrators with absolute control over the AI landscape within their organization. You can precisely authorise or restrict specific AI engines and providers to align with corporate data policies and budgetary goals.

Administering Model Availability

  1. Access the Model Governance section within the Enterprise Portal.
  2. Utilise the selection matrix to enable or disable specific providers and individual models.
  3. Select "Authorize Changes" to propagate settings across the department.

Coming Soon !!!

Advanced Governance Filters

Engineers can only access models that cross-reference with your established filters:

Strategic FilterGovernance Utility
Compliance TierAuthorise only those models that adhere to specific data retention and privacy mandates.
Provider GeographyRestrict AI traffic to specific geographic regions to meet sovereignty requirements.
Model GenerationStandardize on specific families (e.g., Claude 3.5, GPT-4o, Gemini 2.0).
Authorized ProvidersWhitelist entire providers like Anthropic, Google, or OpenAI.
Capability MatrixFilter by model modalities (Code, Vision, High-Context, etc.).
Cost EfficiencyCompare and restrict models based on token expenditure tiers.

Strategic Use Cases

  • Privacy-First Environments: Automatically disable models that retain user prompts or operate outside authorized jurisdictions.
  • Financial Oversight: Limit access to high-cost reasoning models for routine development tasks.
  • Mission Specificity: Enable specialized code-centric engines while disabling general-purpose creative ones.

Critical Policy Notes

  • Privilege Mandate: Only Administrators and Owners can calibrate model governance settings.
  • Rapid Propagation: Governance updates are pushed to all developer environments within seconds.
  • Centralized Enforcement: Organization-level restrictions always override individual user preferences.

On this page