Skip to main content
MyAi exposes a set of system tools (the default_api) that serve as the building blocks for everything the platform does. Whether you’re building a custom integration, automating a workflow, or creating an interactive dashboard, these tools are what make it happen.
This section is primarily for Technical IT Users and Pro Users who want to understand what’s happening under the hood. End Users interact with these tools indirectly through conversation, Canvases, and Workflows.

​
The Tool Landscape

Functions

Deploy custom Python code to orchestrate tools, transform data, and implement business logic.

Data & Integration Tools

Connect to REST APIs, query databases, and ingest files from external systems.

Search & Metadata

Discover artifacts programmatically and manage the relationships between system components.

UI Tools

Build dashboards, interactive apps, and manage the audit trail.

​
How Tools Work Together

A typical MyAi automation chains multiple tools in sequence:
1

Ingest

Use file_processor or api_client to pull data from external systems into MyAi.
2

Transform

Use functions to apply business logic, clean data, or orchestrate multiple tool calls.
3

Store & Link

Use search_artifacts to find related data and relationships to connect components in the Context Graph.
4

Present

Use canvas_editor or app to display results in dashboards, reports, or interactive UIs.
5

Audit

Every step is automatically recorded as a Work Order, providing full traceability via work_orders.

​
Key Concept: Execution References

Execution References (exec_*) are portable handles to data payloads returned by tools like sql_client or api_client.
  • Immutable — once a tool returns a result, the reference is locked.
  • No re-serialization — pass a ref_id directly to a canvas_editor or email tool. The system resolves the data at render time, ensuring data integrity and preventing truncation.
This means you can query a database, get a reference, and hand it to a Canvas chart — without ever needing to copy or reformat the data.

​
Security Model

Never hardcode credentials in function code. Always use integration_credentials artifacts and pass their IDs as parameters.
All external connections are authenticated through Integration Credentials — encrypted artifacts scoped by permissions. This ensures:
  • Credentials are stored securely, not in plaintext code
  • Access is controlled per Dimension and per user role
  • Credential rotation happens in one place, not across every function

​
Learn More

Integrations

See the full connector catalog and supported systems.

Architecture

Understand how Dimensions, Skills, and Artifacts fit together.