​

Schema Naming

• Use descriptive names: Choose names that clearly indicate the schema’s purpose or the type of document it handles (e.g., standard_supplier_invoice, monthly_bank_statement_chase).
• Include version information: If you anticipate schema changes, incorporate versioning in the name (e.g., invoice_schema_v2, statement_fields_2024_q1).
• Keep names consistent: Establish a naming convention and apply it uniformly across your organization for clarity.

​

Schema Versioning

• Automatic Versioning: The Invaro API automatically handles versioning. Each time you update a schema using the PUT /api/v1/schemas/id/{schema_id} endpoint, the version number is automatically incremented.
• History: Previous versions are maintained, allowing you to track changes over time, although only the latest active version is used for processing.

​

Schema Types

• Supported Types: Currently, the API supports "invoice" and "bank_statement" types.
• Choose Correctly: Select the type that matches the documents you intend to process with this schema. This helps the system apply the correct underlying models.
• Immutable Type: The type of a schema cannot be changed after it has been created.

​

Schema String Format (schema_string)

• Define Fields: Clearly list all the data fields you expect to extract within the fields array.
• Consistency: While the AI is flexible, maintaining a consistent structure for similar document types across schemas can be beneficial.
• Use Description: Utilize the main schema description field to document the purpose of specific fields within the schema_string if needed.

​

Schema Management Workflow

• Review Regularly: Periodically review your active schemas to ensure they are still relevant and accurate for the documents you are processing.
• Update as Needed: Modify schemas when the format of your source documents changes or when you need to extract additional fields.
• Test Thoroughly: Before relying on a new or updated schema in production, test it with a variety of sample documents to ensure it extracts data correctly.
• Delete Unused Schemas: Remove schemas that are no longer needed using the DELETE /api/v1/schemas/id/{schema_id} endpoint to keep your schema list manageable.