Secoda AI Capabilities

What Secoda AI has access to answer questions about

The following document outlines what Secoda AI has access to and can answer questions about and what it currently cannot answer. The team at Secoda is quickly evolving the Secoda AI capabilities which will be reflected in this document.

In scope

This section provides an overview of the capabilities of Secoda AI in the areas of data discovery, cataloging, analysis, querying, and metadata governance. It details what Secoda AI can access, including searching for and retrieving data assets and documentation, executing queries, generating visualizations, and handling metadata.

Data Discovery & Catalog

• Search resources - Find and discover data assets (tables, views, dashboards, etc.)

• Search knowledge - Access documents, definitions, and documentation

• Retrieve resource details - Get metadata, schemas, descriptions, and properties of data assets

• Resource links - Generate links to resources within Secoda

• Lineage information - Retrieve upstream and downstream data lineage

• Glossary terms - Access business glossary definitions and terms

Data Analysis & Querying

• Run SQL queries - Execute SQL queries on connected data sources (can be disabled by admins)

• File operations - Read file contents, view first/last lines, search within files

• Chart generation - Create visualizations and charts from data

• Query usage insights – Analyze patterns like most-run queries, user activity, and table access.

• Slow/expensive query detection – Identify and describe costly queries when performance metadata is accessible.

Metadata & Governance

• Glossary & documentation suggestions – Recommend new terms, definitions, and resource documentation based on metadata patterns.

• Monitor suggestions – Proposes monitors based on common incident patterns and gaps in current coverage.

• Tags and owners - Access information about data tags and resource owners

• Incident patterns – Surface historical data quality issues and aggregate incident trends, when sufficient monitor data exists.

• Test visibility – Summarize test coverage levels for key assets

• Integrations - Information about connected data sources

• Teams and users - Access to workspace member information

Knowledge Management

• Documentation - Access to Secoda documentation and help content

• Memory functions - Store and retrieve information for context in conversations

• Note-taking - Create and read notes during research sessions

External Information

• Web search - Search the internet for additional context (when enabled)

• Web page reading - Read content from web pages

Out of scope

This section outlines the limitations and capabilities across various features within the platform, providing clarity on what functionalities are accessible under different categories such as Knowledge Management, External Information, Administrative & Security Features, and System Administration.

Administrative & Security Features

• Policies - Data governance policies, compliance rules, and policy configurations

• Access Requests - Access request workflows, approvals, and permission requests

• Monitors & Incidents - Data quality monitors, incident reports, and monitoring configurations

• Audit logs - Workspace activity history and user action records

System Administration

• User management - User account creation, modification, and deletion

• Workspace settings - Workspace configurations, billing information, and administrative settings

• Integration configurations - Data source connection setup and configuration changes

• Permission management - Role assignments and user permission modifications

Monitoring & Testing (Partial Support)

• Monitor editing – AI can suggest monitors, but not create, activate, or modify live monitors directly.

• Test execution data – Limited ability to analyze test pass/fail rates or show test run history unless metadata is present.

Automation & System Controls

• Automation configurations – AI can describe automations conceptually, but cannot currently access or modify scheduled jobs or execution configs.

• Automation health insights – AI cannot yet identify failed automation jobs or suggest workflow improvements.

• Integration setup – Cannot configure data source connections or modify integration credentials.

Access Controls & Restrictions

Access controls in user management, automation, and AI are crucial for securing resources and safeguarding data privacy. Users can configure access levels through persona-based filtering and control features in alignment with:

Persona-Based Filtering

• AI Personas can have included/excluded filters that restrict which resources the AI can access

• Admins can configure governance rules to limit AI access to specific data assets

Tool-Level Controls

• Individual AI tools can be enabled or disabled by workspace administrators

• By default, some tools like SQL execution may be disabled for security

Data Privacy

• PII masking - Can be configured to mask personally identifiable information

• Entity permissions - Respects user-level permissions for data access

• Team-based access - AI access follows the same team-based permissions as users

Feature Access

• Access depends on subscription plan and feature flags

• Some features may be in beta or require specific workspace configurations