Configuration guide

Admin setup & configuration guide

For Saidot organisation admins

This comprehensive guide provides an overview of all administrative setup and configuration tasks available in the Saidot platform. Use this as your central reference point for configuring your organisation's AI governance environment. Typically configuration is done in collaboration with Saidot Customer Success team.

Getting started

Before diving into specific configurations, familiarize yourself with the Saidot platform and its core concepts.

📖 Essential reading:

• Getting Started with Saidot - Platform overview and key features

• Introduction to Saidot Methodology - AI governance framework and best practices

1. Authentication

1.1. Activate Admin account

As an identified admin, you will receive an invitation to your email to activate your account. When you have received the invitation, verify your email address and activate your account by clicking on the verification button or the link in the email. The Sign up-activation screen requires your to add your name, title and user group(s) and setting a password. In case of forgetting your password after activation has been completed, you can reset it easily by selecting Forgot password? in the login screen and entering your email for receiving a password reset link.

📖 Detailed Guide: Login to Saidot

1.2. Authentication configuration

Configure how users access your Saidot organisation. Choose between native Saidot accounts or integrate with your existing identity provider. General recommendation for both cases is to:

• Define internal responsibilities for access rights management

• Create internal guideline for access requests

• Agree ways how users are informed of their access and Saidot use

Available options:

• Saidot Native Accounts - Email-based authentication with native accounts management

• Microsoft Entra ID SSO - Single sign-on integration for seamless access

Please note that when using Entra ID, the Space specific access rights are still managed in Saidot Admin Settings. The users who have been added to and Entra ID Group, will not receive Saidot invitation emails or product release newsletters via email.

1.2.1. Admin tasks for native accounts

• Send invitation emails to new members

• Manage user activation and deactivation

• Control password resets and access recovery

• View and manage all organisation members and Space level access

1.2.2. Admin tasks for Entra ID authentication

• Enable the Entra ID toggle in Organisation Settings

• Register the Saidot application in Entra

• Create Saidot Admin and Member user groups in Entra

• Map your Entra tenancy and user groups to Saidot

• Agree and create a process to grant access rights and send user information

📖 Detailed guides: Login to Saidot and How to manage members and spaces

2. Spaces & access control

Organise your AI governance work into separate Spaces with granular access control. Spaces are a solution for large corporations with independent business units, who want separate AI inventories and catalogues. They can also be used to isolate confidential AI systems from the General space that all the units have access to.

Key concepts:

• General Space - Default space for all organisation members

• Custom Spaces - Create dedicated workspaces for business units or specific AI systems

• Space Managers - Delegate space-level administration

Admin tasks:

• Create and configure spaces

• Assign Space Managers

• Control AI system visibility across spaces

• Manage space-specific member access

📖 Detailed guide: How to manage members and spaces

3. Roles & responsibilities

Decide how to assign governance roles using the Three Lines of Defence model. Admins can edit and see all AI systems, but members can only edit the AI systems whose teams they are invited to. The Members who are invited to the System team as Reviewers can be assigned Review tasks.

Available roles:

• System Owner - Overall accountability for AI system governance

• Business Owner - Business perspective and requirements

• Risk Owner - Risk management and mitigation oversight

• Model owner - Model related governance

• Data owner - Data related governance

• Reviewer - Reviewer and approved of the system

• Oversight manager - Managing overall AI governance progress

Admin capabilities:

• Assign roles at System level

• Remove members from the AI system teams

• Add missing members to the AI system teams

📖 Detailed guides:

• Assign roles and responsibilities

• How to manage members and spaces

4. API & Integration management

Saidot allows you to build integrated AI governance workflows through native in-built integrations to AI/ML development platforms, through our extensive selection of API’s or MCP servers.

4.1. Saidot integrations

Configure integrations that are supported by Saidot natively. The integration capabilities are expanded on an active basis. Check the latest status from Release notes or with your Saidot Customer Success team.

Available integrations:

• Microsoft Entra ID (authentication)

• Azure ML/AI Services (Model Catalogue)

Admin responsibilities:

• Configure integration credentials

• Monitor integration health

• Troubleshoot connection issues

• Manage data sync schedules

📖 Detailed guides: Using Model Catalogue

4.2. Saidot API’s

Create and manage API tokens for programmatic access to Saidot.

Admin tasks:

• Generate new API tokens

• Set token expiration dates

• Revoke and regenerate tokens

• Store tokens securely

Exmple use cases:

• Automated system registration

• Bulk data imports for Catalogues or AI Systems

• Custom integrations for AI development, GRC and Data Catalog tools

• CI/CD pipeline integration

📖 Detailed guide: Working with Saidot APIs

🔗 API documentation: Access Swagger docs through the Admin interface or Release Notes

5. Model Catalogue setup

Model catalogue setup starts by connecting to your AI/ML development platforms. Saidot allows you to use built-in integrations for Azure and AWS Services. In addition, customers may use Model Catalogue API to integrate the model metadata from other platforms that are not supported by in-built integrations-

Key features:

• Automatic model metadata import

• Model governance, including ownership, approvals and accepted use policy

• Risk inheritance for base models from Model Library

• Automated generation of Evaluation plans (Azure)

📖 Detailed guide: Using Model Catalogue

5.1. Azure integration

Connect your Azure ML and AI services to automatically import and govern models. Supported Azure Services are Azure Machine Learning, Azure AI Services and Azure OpenAI Services.

Setup requirements:

1. Register Saidot app in Azure

2. Assign reader roles to Saidot

3. Configure tenant mapping

4. Enable Model Catalogue in Saidot

5.2. Model governance

Once integrated, engage your AI developers to start governing your imported models with Saidot.

Admin capabilities:

• View imported model metadata

• Configure model-to-system mappings

• Enable automatic risk inheritance

• Enable automated model evaluations plan creation

📖 Detailed guides: Using Model Catalogue and Working on system overview

6. Automations

Configure automated governance workflows to streamline compliance activities.

Available automations:

• AI Act classification - Automatic risk classification

• Risk inheritance - Auto-propagate risks from datasets/models

• Risk level inheritance - Inherit general risk level from system components Upcoming feature

Admin capabilities:

• Enable/disable automations at space level

• Configure automation triggers

• Monitor automation executions

• Track time savings and efficiency gains

📖 Detailed guide: Automate governance workflows

7. Custom settings and fields

Configure default settings and custom fields and dimensions for new AI system registrations.

7.1. Organisation settings

Configure your organisation's basic information and branding.

Configuration options:

• Organisation logo, name and details

• Manage suggested values for region, industry and business unit

📖 Detailed guide: How to manage members and spaces

7.2. Custom fields

Use custom fields to link AI Systems to a specific customer, supplier, product or process.

Configuration options:

• Download values for Customers

• Download values for Products

• Download values for Suppliers

📖 Detailed guide: Adding custom lists

8. Custom assessments

Extend the platform with organisation-specific policies.

Configuration options:

• Create custom assessment (only available for Saidot Admins, please contact customersupport@saidot.ai)

9. Evidence store activation

Activate Evidence store in Organisation Settings. You may restrict or allow Evidence store to use AI-based recommendations or share evidence across Spaces. AI-enabled Evidence Store can be governed at Saidot as an AI System using the Evidence Story Product Card in the Product Library.

📖 Detailed guide: Using Evidence store

10. Dataset Catalogue setup

Configure dataset governance and tracking.

Admin capabilities:

• Decide the ways to populate Data Catalogue: Manual entry or through API

• Use API to integrate Datasets to Saidot Data Catalogue from external Data Catalog Tool or Data Governance Platform

📖 Detailed guide: How to manage and link datasets and Working with Saidot APIs

11. Reviews

Define how AI systems are reviewed.

Admin capabilities:

• Create instructions for AI system reviews

• Define entities and roles that are allowed to act as reviewers

• Instruct the users to create and assign Review tasks

• Instruct the Reviewers to log in audit reports and decisions to Reviews

📖 Related guide: Review systems

12. Reporting

Monitor platform usage and adoption across your organisation using the Saidot Home page. Admins can also use Saidot API’s for building data pipelines for custom BI reporting.

Available metrics:

13. Support

Saidot Customer Success team is assigned to help your admins in configurations, first AI systems registrations and setting up the needed integrations. The Saidot Customer Support, AI Governance experts and the Library team is available to assist in any technical issues, Library content request or developing your AI governance processes and practices with Saidot.

Support channels:

• Personal support: Your assigned Customer Success team

• Email support: Contact through Support page

• Documentation: This knowledge base

• Release Notes: Stay updated on new features and fixes from our Release notes

Before contacting support:

1. Check relevant documentation section

2. Review recent release notes for known issues

3. Verify user permissions and access levels

4. Document error messages and steps to reproduce

14. Best practices and checklists

14.1 Initial setup checklist

Recommended order for a new organisation setup:

1. ✅ Configure authentication (Entra ID SSO recommended)

2. ✅ Set up organisation settings and custom fields if relevant

3. ✅ Create Spaces for restricted access and assign Space Managers

4. ✅ Design integration architecture and decide how to use API’s

5. ✅ Configure Model Catalogue integration (if using Azure)

6. ✅ Enable automation workflows (AI Act classification)

7. ✅ Train Space Managers and key users

8. ✅ Begin system registration process

14.2 Ongoing administration

Regular admin tasks:

• Review and approve new member requests and assign Space level access

• Analyse AI inventory and Catalogue development progress

• Measure AI governance process, efficiency and impact to AI time-to-market

• Monitor integration status and extend integrations when needed

• Review automation execution logs and time saved with automated workflows

• Request updates to Library content if necessary

• Manage API token lifecycle

• Stay current with platform updates and release notes

• Share information to the new Members on how to utilise new features

• Collaborate with Saidot Customer Success Team to align with best practices

14.3 Security recommendations

Access control:

• Use Entra ID SSO for centralized identity management

• Implement least privilege access principles

• Regularly review Space Manager role assignments

• Audit role assignments quarterly

API security:

• Rotate API tokens regularly

• Set appropriate token expiration periods

• Store tokens in secure credential management systems

• Monitor API usage for anomalies

Data governance:

• Configure appropriate data retention policies

• Ensure compliance with regional data regulations

• Regular backup and export of critical governance data

• Document disaster recovery procedures

15. Staying current

15.1 Platform updates

Stay informed about new features and improvements:

📖 Release notes: Release Notes

15.2 Continuous learning

Recommended resources:

• Review new documentation releases

• Use Saidot materials, such as AI Governance Handbook

• Agree on additional Saidot support services and trainings with the Customer Success team

• Provide feedback on platform improvements

Quick reference matrix

Glossary

Organisation admin - User with full administrative privileges across the entire Saidot organisation

Space manager - User with administrative privileges within a specific space

System owner - Person accountable for overall governance of an AI system

Model catalogue - Feature for importing and governing ML/AI models from Azure

API token - Credential for programmatic access to Saidot APIs

Risk inheritance - Automatic propagation of risks from components to systems

Three lines of defence - Governance model separating ownership, oversight, and assurance

Document information

Last Updated: January 2026
Version: 1.0
Maintained By: Saidot Customer Success
Feedback: Contact support

Need Help? If you can't find what you're looking for in this guide, please visit our Support page or contact our customer success team.