Create example set
POST /example-sets
Authorizations
Section titled “Authorizations ”Request Body
Section titled “Request Body ”object
ID of the team that will own this example set
Human-readable name for the example set (1-255 characters)
Optional description of what this example set is for
Schema defining what inputs examples should have
object
File inputs/outputs
object
Unique identifier for this file slot
Human-readable label for the file (e.g., ‘Input Document’)
Optional description of what this file represents
Accepted file types (MIME types or extensions, e.g., ‘.pdf,.docx’)
Whether this file is required
Whether multiple files can be selected for this slot
How file content is retrieved: ‘semantic’ uses vector similarity search, ‘full’ returns all chunks
Dataset inputs/outputs
object
Unique identifier for this dataset slot
Human-readable label for the dataset (e.g., ‘Training Data’)
Optional description of what this dataset represents
JSON Schema describing the expected dataset structure
Whether this dataset is required
Whether multiple datasets can be selected for this slot
Optional row-level filter applied to any dataset in this slot
object
Column name in the dataset table
object
Reference to an input value (e.g., ‘input.dataInputs.tenantId’)
Structured data inputs/outputs
object
Unique identifier for this data input slot
Human-readable label for the data input (e.g., ‘Parameters’)
Optional description of what this data input represents
JSON Schema describing the expected data structure
Whether this data input is required
Schema defining what outputs examples should have
object
File inputs/outputs
object
Unique identifier for this file slot
Human-readable label for the file (e.g., ‘Input Document’)
Optional description of what this file represents
Accepted file types (MIME types or extensions, e.g., ‘.pdf,.docx’)
Whether this file is required
Whether multiple files can be selected for this slot
How file content is retrieved: ‘semantic’ uses vector similarity search, ‘full’ returns all chunks
Dataset inputs/outputs
object
Unique identifier for this dataset slot
Human-readable label for the dataset (e.g., ‘Training Data’)
Optional description of what this dataset represents
JSON Schema describing the expected dataset structure
Whether this dataset is required
Whether multiple datasets can be selected for this slot
Optional row-level filter applied to any dataset in this slot
object
Column name in the dataset table
object
Reference to an input value (e.g., ‘input.dataInputs.tenantId’)
Structured data inputs/outputs
object
Unique identifier for this data input slot
Human-readable label for the data input (e.g., ‘Parameters’)
Optional description of what this data input represents
JSON Schema describing the expected data structure
Whether this data input is required
Optional pre-filled configuration values
object
File configurations
object
Unique identifier for this file slot
Human-readable label for the file (e.g., ‘Input Document’)
Optional description of what this file represents
Accepted file types (MIME types or extensions, e.g., ‘.pdf,.docx’)
Whether this file is required
Whether multiple files can be selected for this slot
How file content is retrieved: ‘semantic’ uses vector similarity search, ‘full’ returns all chunks
Dataset configurations
object
Unique identifier for this dataset slot
Human-readable label for the dataset (e.g., ‘Training Data’)
Optional description of what this dataset represents
JSON Schema describing the expected dataset structure
Whether this dataset is required
Whether multiple datasets can be selected for this slot
Optional row-level filter applied to any dataset in this slot
object
Column name in the dataset table
object
Reference to an input value (e.g., ‘input.dataInputs.tenantId’)
Structured data configurations
object
Unique identifier for this data input slot
Human-readable label for the data input (e.g., ‘Parameters’)
Optional description of what this data input represents
JSON Schema describing the expected data structure
Whether this data input is required
object
Structured data value matching the schema
LLM handler configuration
object
LLM model in {provider}/{model} format (e.g., vertexai/gemini-3-flash-preview)
Generation temperature (0 = deterministic, 2 = creative)
Maximum output tokens
Web search configuration for automatic context retrieval
object
Enable automatic web search query generation and execution
Maximum number of queries to generate (1-5, default: 3)
Maximum results per query (1-20, default: 5)
Search depth: basic (1 credit) or advanced (2 credits, default: basic)
Include AI-generated answer summary from Tavily (default: false)
URL scraping configuration
object
Enable automatic URL detection and scraping (default: true)
Content type for scraped pages (default: ‘markdown’)
Enable JavaScript rendering (default: true)
Maximum planning/retrieval iterations. Each iteration plans what to retrieve, fetches context, and evaluates sufficiency before answering.
Responses
Section titled “ Responses ”Example set created successfully
object
Unique identifier for the example set (nanoid format)
ID of the team that owns this example set
Human-readable name for the example set
Optional description of what this example set is for
ID of the currently active configuration version
Schema defining what inputs examples should have
object
File inputs/outputs
object
Unique identifier for this file slot
Human-readable label for the file (e.g., ‘Input Document’)
Optional description of what this file represents
Accepted file types (MIME types or extensions, e.g., ‘.pdf,.docx’)
Whether this file is required
Whether multiple files can be selected for this slot
How file content is retrieved: ‘semantic’ uses vector similarity search, ‘full’ returns all chunks
Dataset inputs/outputs
object
Unique identifier for this dataset slot
Human-readable label for the dataset (e.g., ‘Training Data’)
Optional description of what this dataset represents
JSON Schema describing the expected dataset structure
Whether this dataset is required
Whether multiple datasets can be selected for this slot
Optional row-level filter applied to any dataset in this slot
object
Column name in the dataset table
object
Reference to an input value (e.g., ‘input.dataInputs.tenantId’)
Structured data inputs/outputs
object
Unique identifier for this data input slot
Human-readable label for the data input (e.g., ‘Parameters’)
Optional description of what this data input represents
JSON Schema describing the expected data structure
Whether this data input is required
Schema defining what outputs examples should have
object
File inputs/outputs
object
Unique identifier for this file slot
Human-readable label for the file (e.g., ‘Input Document’)
Optional description of what this file represents
Accepted file types (MIME types or extensions, e.g., ‘.pdf,.docx’)
Whether this file is required
Whether multiple files can be selected for this slot
How file content is retrieved: ‘semantic’ uses vector similarity search, ‘full’ returns all chunks
Dataset inputs/outputs
object
Unique identifier for this dataset slot
Human-readable label for the dataset (e.g., ‘Training Data’)
Optional description of what this dataset represents
JSON Schema describing the expected dataset structure
Whether this dataset is required
Whether multiple datasets can be selected for this slot
Optional row-level filter applied to any dataset in this slot
object
Column name in the dataset table
object
Reference to an input value (e.g., ‘input.dataInputs.tenantId’)
Structured data inputs/outputs
object
Unique identifier for this data input slot
Human-readable label for the data input (e.g., ‘Parameters’)
Optional description of what this data input represents
JSON Schema describing the expected data structure
Whether this data input is required
Pre-filled configuration values
object
File configurations
object
Unique identifier for this file slot
Human-readable label for the file (e.g., ‘Input Document’)
Optional description of what this file represents
Accepted file types (MIME types or extensions, e.g., ‘.pdf,.docx’)
Whether this file is required
Whether multiple files can be selected for this slot
How file content is retrieved: ‘semantic’ uses vector similarity search, ‘full’ returns all chunks
Dataset configurations
object
Unique identifier for this dataset slot
Human-readable label for the dataset (e.g., ‘Training Data’)
Optional description of what this dataset represents
JSON Schema describing the expected dataset structure
Whether this dataset is required
Whether multiple datasets can be selected for this slot
Optional row-level filter applied to any dataset in this slot
object
Column name in the dataset table
object
Reference to an input value (e.g., ‘input.dataInputs.tenantId’)
Structured data configurations
object
Unique identifier for this data input slot
Human-readable label for the data input (e.g., ‘Parameters’)
Optional description of what this data input represents
JSON Schema describing the expected data structure
Whether this data input is required
object
Structured data value matching the schema
LLM handler configuration
object
LLM model in {provider}/{model} format (e.g., vertexai/gemini-3-flash-preview)
Generation temperature (0 = deterministic, 2 = creative)
Maximum output tokens
Web search configuration for automatic context retrieval
object
Enable automatic web search query generation and execution
Maximum number of queries to generate (1-5, default: 3)
Maximum results per query (1-20, default: 5)
Search depth: basic (1 credit) or advanced (2 credits, default: basic)
Include AI-generated answer summary from Tavily (default: false)
URL scraping configuration
object
Enable automatic URL detection and scraping (default: true)
Content type for scraped pages (default: ‘markdown’)
Enable JavaScript rendering (default: true)
Maximum planning/retrieval iterations. Each iteration plans what to retrieve, fetches context, and evaluates sufficiency before answering.
Example set status: ‘active’ or ‘archived’
When the example set was created (ISO 8601)
When the example set was last modified (ISO 8601)
ID of the user who created this example set
Bad Request - Validation error or invalid input
object
Unauthorized - Authentication required or invalid token
object
Forbidden - Insufficient permissions