Skip to main content

Examples and Instructions for Context API v2

This guide provides comprehensive examples for using the Context API v2 with detailed explanations of action-specific instructions.

Overview​

The Context API v2 introduces structured actions where:

  • actions is an object (not an array)
  • Action names are keys in camelCase format
  • Each action's value is an object containing its configuration parameters

This allows for:

  • Independent configuration per action
  • Type-safe parameter passing
  • Clear validation messages
  • Better extensibility

Configuration Parameters vs Instructions​

Each action has configuration parameters that control its behavior (e.g., maxWordCount, classes, kSimilarMetadata).

Additionally, actions that support an optional instructions field for custom AI guidance:

  • textSummarization - Guide summarization style and focus
  • imageDescription - Control description format and detail level
  • namedEntityRecognitionText - Specify entity types and focus areas
  • namedEntityRecognitionImage - Guide OCR and entity extraction
  • textClassification - Provide context for classification decisions
  • imageClassification - Specify classification focus areas
  • textMetadataGeneration - Guide metadata extraction strategy
  • imageMetadataGeneration - Control metadata identification approach

The instructions field accepts a custom object to provide additional context or guidance to the AI model.

Action Categories​

Text Processing Actions​

Text Summarization​

Generate concise summaries of text documents with optional AI guidance.

Action Name: textSummarization

Configuration Parameters:

  • maxWordCount (optional): Maximum number of words in the summary
  • instructions (optional): Custom instructions object for summarization guidance

Example - Basic:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/report.pdf"
    }
  ],
  "actions": {
    "textSummarization": {}
  }
}

Example - With Word Limit:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/long-report.pdf"
    }
  ],
  "actions": {
    "textSummarization": {
      "maxWordCount": 150
    }
  }
}

Example - Financial Report with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/quarterly-earnings.pdf"
    }
  ],
  "actions": {
    "textSummarization": {
      "maxWordCount": 200,
      "instructions": {
        "focus": "Highlight key financial metrics, revenue, and outlook",
        "style": "professional",
        "emphasize": "financial_performance|growth_trends"
      }
    }
  }
}

Example - Legal Document with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/contract-terms.pdf"
    }
  ],
  "actions": {
    "textSummarization": {
      "maxWordCount": 250,
      "instructions": {
        "focus": "Extract key obligations, terms, conditions, and deadlines",
        "style": "structured",
        "highlightRisks": true,
        "context": "legal contract analysis"
      }
    }
  }
}

Example - Medical Document with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/patient-report.pdf"
    }
  ],
  "actions": {
    "textSummarization": {
      "maxWordCount": 200,
      "instructions": {
        "focus": "Summarize diagnosis, treatment plan, and follow-up instructions",
        "style": "clinical",
        "emphasize": "critical_findings|treatment_recommendations",
        "context": "medical report summary"
      }
    }
  }
}

Output:

{
  "id": "abc123",
  "results": [
    {
      "objectKey": "documents/report.pdf",
      "textSummary": {
        "isSuccess": true,
        "result": "This report analyzes quarterly performance...",
        "error": null
      }
    }
  ]
}

Text Metadata Generation​

Extract structured metadata from text using AI-powered analysis with example-based learning.

Action Name: textMetadataGeneration

Configuration Parameters:

  • kSimilarMetadata (required): Array of metadata example objects showing desired metadata structure
  • instructions (optional): Custom instructions object for AI guidance

Metadata Format Guidelines:

  • Use : to separate category from field name (e.g., "document:type")
  • Use | to separate possible values (e.g., "Invoice|Receipt|Contract")
  • Provide representative examples of expected values
  • Multiple metadata objects help the AI understand variations

Example - Document Classification:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/contract.pdf"
    }
  ],
  "actions": {
    "textMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "document:type": "Contract|Agreement|NDA|MOU",
          "document:category": "Legal|Financial|Employment|Corporate",
          "party:count": "2|3|Multiple",
          "keywords:tags": "confidentiality|partnership|employment|services"
        }
      ]
    }
  }
}

Example - Meeting Minutes Metadata:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/meeting-notes.pdf"
    }
  ],
  "actions": {
    "textMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "meeting:type": "Board Meeting|Team Meeting|Client Call|Review",
          "attendees:count": "2-5|6-10|11+",
          "has:action_items": "yes|no",
          "has:decisions": "yes|no",
          "topics:list": "budget|strategy|project_update|personnel",
          "urgency:level": "high|medium|low"
        }
      ]
    }
  }
}

Example - Technical Documentation:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/api-spec.pdf"
    }
  ],
  "actions": {
    "textMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "document:type": "API Documentation|User Manual|Technical Spec|Guide",
          "technology:stack": "REST|GraphQL|gRPC|SOAP",
          "programming:language": "Python|JavaScript|Java|C#|Go",
          "version:type": "v1|v2|v3|beta|alpha",
          "complexity:level": "beginner|intermediate|advanced",
          "has:code_examples": "yes|no",
          "keywords:tags": "authentication|endpoints|webhooks|integration"
        }
      ]
    }
  }
}

Example - With Custom Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/legal-agreement.pdf"
    }
  ],
  "actions": {
    "textMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "document:type": "Contract|Agreement|NDA|MOU",
          "document:category": "Legal|Financial|Employment|Corporate",
          "keywords:tags": "confidentiality|partnership|employment|services"
        }
      ],
      "instructions": {
        "focus": "Extract formal legal terminology",
        "detailLevel": "high"
      }
    }
  }
}

Output:

{
  "id": "def456",
  "results": [
    {
      "objectKey": "documents/contract.pdf",
      "textMetadata": {
        "isSuccess": true,
        "result": {
          "document:type": "NDA",
          "document:category": "Legal",
          "party:count": "2",
          "keywords:tags": "confidentiality"
        },
        "error": null
      }
    }
  ]
}

Text Classification​

Classify documents into predefined categories.

Action Name: textClassification

Configuration Parameters:

  • classes (required): Array of at least 2 distinct classification categories
  • instructions (optional): Custom instructions object for AI guidance

Example - Document Type Classification:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/invoice-2024.pdf"
    }
  ],
  "actions": {
    "textClassification": {
      "classes": [
        "invoice",
        "receipt",
        "contract",
        "report",
        "correspondence",
        "other"
      ]
    }
  }
}

Example - Sentiment Classification:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/customer-feedback.pdf"
    }
  ],
  "actions": {
    "textClassification": {
      "classes": [
        "positive",
        "negative",
        "neutral"
      ]
    }
  }
}

Example - Industry Classification:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/business-plan.pdf"
    }
  ],
  "actions": {
    "textClassification": {
      "classes": [
        "technology",
        "healthcare",
        "finance",
        "retail",
        "manufacturing",
        "services"
      ]
    }
  }
}

Example - With Custom Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/legal-contract.pdf"
    }
  ],
  "actions": {
    "textClassification": {
      "classes": [
        "invoice",
        "contract",
        "receipt",
        "report",
        "other"
      ],
      "instructions": {
        "context": "legal and financial documents",
        "priority": "accuracy over speed"
      }
    }
  }
}

Output:

{
  "id": "ghi789",
  "results": [
    {
      "objectKey": "documents/invoice-2024.pdf",
      "textClassification": {
        "isSuccess": true,
        "result": "invoice",
        "error": null
      }
    }
  ]
}

Text Embeddings​

Generate vector embeddings for semantic search and similarity matching.

Action Name: textEmbeddings

Configuration Parameters: None required

⚠️ Breaking Change in v2: The output format for text embeddings has changed from nested array List<List<float>> to flat array List<float>. See Breaking Changes to the Embeddings Format for migration details.

Example:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/article.pdf"
    }
  ],
  "actions": {
    "textEmbeddings": {}
  }
}

Output (v2 - Flat Array):

{
  "id": "jkl012",
  "results": [
    {
      "objectKey": "documents/article.pdf",
      "textEmbeddings": {
        "isSuccess": true,
        "result": [-0.021741372, 0.005967313, -0.02682978, 0.01471012],
        "error": null
      }
    }
  ]
}

Named Entity Recognition (Text)​

Extract named entities (people, organizations, locations, etc.) from text documents.

Action Name: namedEntityRecognitionText

Configuration Parameters:

  • instructions (optional): Custom instructions object for entity extraction guidance

Example - Basic:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/news-article.pdf"
    }
  ],
  "actions": {
    "namedEntityRecognitionText": {}
  }
}

Example - Legal Document with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/contract.pdf"
    }
  ],
  "actions": {
    "namedEntityRecognitionText": {
      "instructions": {
        "focus": "Extract party names and signatory individuals",
        "entityPriority": "PERSON,ORGANIZATION",
        "includeRoles": true,
        "context": "legal document with multiple parties"
      }
    }
  }
}

Example - Medical Document with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/medical-report.pdf"
    }
  ],
  "actions": {
    "namedEntityRecognitionText": {
      "instructions": {
        "focus": "Extract patient names, doctor names, hospital names, and medication names",
        "customEntityTypes": "PERSON,ORGANIZATION,MEDICATION",
        "context": "medical report with patient and provider information"
      }
    }
  }
}

Example - Business Email with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/business-correspondence.txt"
    }
  ],
  "actions": {
    "namedEntityRecognitionText": {
      "instructions": {
        "focus": "Extract sender, recipient, and mentioned companies",
        "prioritizeEmailAddresses": true,
        "context": "business communication with contact information"
      }
    }
  }
}

Output:

{
  "id": "mno345",
  "results": [
    {
      "objectKey": "documents/news-article.pdf",
      "namedEntityText": {
        "isSuccess": true,
        "result": {
          "PERSON": ["John Smith", "Jane Doe"],
          "ORGANIZATION": ["Microsoft", "Google"],
          "LOCATION": ["New York", "California"]
        },
        "error": null
      }
    }
  ]
}

Image Processing Actions​

Image Description​

Generate natural language descriptions of images using computer vision and AI.

Action Name: imageDescription

Configuration Parameters:

  • maxWordCount (optional): Maximum number of words in the description
  • instructions (optional): Custom instructions object for description guidance

Example - Basic:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/product-photo.jpg"
    }
  ],
  "actions": {
    "imageDescription": {}
  }
}

Example - With Word Limit:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/landscape.jpg"
    }
  ],
  "actions": {
    "imageDescription": {
      "maxWordCount": 50
    }
  }
}

Example - Accessibility Description with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/webpage-screenshot.png"
    }
  ],
  "actions": {
    "imageDescription": {
      "maxWordCount": 100,
      "instructions": {
        "focus": "Provide alt-text suitable for accessibility",
        "detailLevel": "comprehensive",
        "includeLayout": true,
        "context": "web accessibility for visually impaired users"
      }
    }
  }
}

Example - Product Catalog with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/product-item.jpg"
    }
  ],
  "actions": {
    "imageDescription": {
      "maxWordCount": 75,
      "instructions": {
        "focus": "Emphasize product features, materials, and dimensions",
        "includeBrandName": true,
        "context": "e-commerce product listing"
      }
    }
  }
}

Example - Medical Imaging with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/xray-scan.jpg"
    }
  ],
  "actions": {
    "imageDescription": {
      "maxWordCount": 150,
      "instructions": {
        "focus": "Technical medical imaging description for clinical use",
        "includeTechnicalTerms": true,
        "detailLevel": "high",
        "context": "medical imaging documentation"
      }
    }
  }
}

Output:

{
  "id": "pqr678",
  "results": [
    {
      "objectKey": "images/product-photo.jpg",
      "imageDescription": {
        "isSuccess": true,
        "result": "A modern smartphone with a sleek black design...",
        "error": null
      }
    }
  ]
}

Image Metadata Generation​

Extract structured metadata from images using AI with example-based learning.

Action Name: imageMetadataGeneration

Configuration Parameters:

  • kSimilarMetadata (required): Array of metadata example objects
  • instructions (optional): Custom instructions object for AI guidance

Example - Product Images:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/clothing-item.jpg"
    }
  ],
  "actions": {
    "imageMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "product:type": "T-Shirt|Dress|Pants|Shoes|Jacket|Hat",
          "product:color": "Red|Blue|Green|Black|White|Yellow|Multi-color",
          "product:gender": "Male|Female|Unisex",
          "product:season": "Summer|Winter|Fall|Spring|All-Season",
          "style:type": "Casual|Formal|Sports|Business|Vintage",
          "keywords:tags": "fashion|clothing|apparel|style"
        }
      ]
    }
  }
}

Example - Vehicle Inspection:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/vehicle-inspection.jpg"
    }
  ],
  "actions": {
    "imageMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "vehicle:type": "Sedan|SUV|Truck|Van|Motorcycle|Bus",
          "vehicle:color": "White|Black|Silver|Red|Blue|Other",
          "vehicle:condition": "Excellent|Good|Fair|Poor|Damaged",
          "damage:visible": "yes|no",
          "damage:location": "Front|Rear|Side|Multiple|None",
          "damage:severity": "Minor|Moderate|Severe|Total",
          "environment:location": "Street|Parking Lot|Driveway|Highway|Indoor"
        }
      ]
    }
  }
}

Example - Real Estate Photos:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/property-listing.jpg"
    }
  ],
  "actions": {
    "imageMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "property:type": "House|Apartment|Condo|Townhouse|Commercial",
          "room:type": "Living Room|Bedroom|Kitchen|Bathroom|Office|Exterior",
          "style:architecture": "Modern|Traditional|Contemporary|Victorian|Colonial",
          "condition:overall": "New|Excellent|Good|Needs Work|Renovation",
          "features:notable": "Fireplace|Pool|Garden|Balcony|Hardwood Floors",
          "lighting:quality": "Bright|Natural|Dim|Well-lit",
          "furnishing:status": "Furnished|Partially Furnished|Unfurnished|Staged"
        }
      ]
    }
  }
}

Example - With Custom Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/product-photo.jpg"
    }
  ],
  "actions": {
    "imageMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "product:type": "T-Shirt|Dress|Pants|Shoes|Jacket|Hat",
          "product:color": "Red|Blue|Green|Black|White|Yellow|Multi-color",
          "style:type": "Casual|Formal|Sports|Business|Vintage",
          "keywords:tags": "fashion|clothing|apparel|style"
        }
      ],
      "instructions": {
        "focus": "Emphasize brand and quality indicators",
        "detailLevel": "high"
      }
    }
  }
}

Output:

{
  "id": "stu901",
  "results": [
    {
      "objectKey": "images/clothing-item.jpg",
      "imageMetadata": {
        "isSuccess": true,
        "result": {
          "product:type": "T-Shirt",
          "product:color": "Blue",
          "product:gender": "Unisex",
          "product:season": "Summer",
          "style:type": "Casual",
          "keywords:tags": "fashion"
        },
        "error": null
      }
    }
  ]
}

Image Classification​

Classify images into predefined categories.

Action Name: imageClassification

Configuration Parameters:

  • classes (required): Array of at least 2 distinct classification categories
  • instructions (optional): Custom instructions object for AI guidance

Example - Vehicle Damage Assessment:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/car-photo.jpg"
    }
  ],
  "actions": {
    "imageClassification": {
      "classes": [
        "damaged_vehicle",
        "undamaged_vehicle",
        "not_a_vehicle"
      ]
    }
  }
}

Example - Document Type from Image:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/scanned-doc.jpg"
    }
  ],
  "actions": {
    "imageClassification": {
      "classes": [
        "passport",
        "drivers_license",
        "id_card",
        "birth_certificate",
        "other_document"
      ]
    }
  }
}

Example - Scene Classification:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/photo.jpg"
    }
  ],
  "actions": {
    "imageClassification": {
      "classes": [
        "indoor",
        "outdoor",
        "nature",
        "urban",
        "portrait"
      ]
    }
  }
}

Example - With Custom Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/vehicle-damage.jpg"
    }
  ],
  "actions": {
    "imageClassification": {
      "classes": [
        "damaged_vehicle",
        "undamaged_vehicle",
        "not_a_vehicle"
      ],
      "instructions": {
        "focus": "Pay attention to minor scratches and dents",
        "context": "insurance claim assessment"
      }
    }
  }
}

Output:

{
  "id": "vwx234",
  "results": [
    {
      "objectKey": "images/car-photo.jpg",
      "imageClassification": {
        "isSuccess": true,
        "result": "damaged_vehicle",
        "error": null
      }
    }
  ]
}

Image Embeddings​

Generate vector embeddings for image similarity search.

Action Name: imageEmbeddings

Configuration Parameters: None required

⚠️ Breaking Change in v2: The output format for image embeddings has changed from nested array List<List<float>> to flat array List<float>. See Breaking Changes to the Embeddings Format for migration details.

Example:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/product.jpg"
    }
  ],
  "actions": {
    "imageEmbeddings": {}
  }
}

Output (v2 - Flat Array):

{
  "id": "yz567",
  "results": [
    {
      "objectKey": "images/product.jpg",
      "imageEmbeddings": {
        "isSuccess": true,
        "result": [0.234, -0.567, 0.891, -0.123, 0.456],
        "error": null
      }
    }
  ]
}

Named Entity Recognition (Image)​

Extract text entities from images using Optical Character Recognition (OCR) combined with Named Entity Recognition.

Action Name: namedEntityRecognitionImage

Configuration Parameters:

  • instructions (optional): Custom instructions object for OCR and entity extraction guidance

Example - Basic:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/business-card.jpg"
    }
  ],
  "actions": {
    "namedEntityRecognitionImage": {}
  }
}

Example - Invoice Document with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/invoice-scan.jpg"
    }
  ],
  "actions": {
    "namedEntityRecognitionImage": {
      "instructions": {
        "focus": "Extract company names, invoice numbers, and monetary amounts",
        "prioritize": "ORGANIZATION,MONEY,DATE",
        "context": "financial document processing"
      }
    }
  }
}

Example - Identification Document with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/drivers-license.jpg"
    }
  ],
  "actions": {
    "namedEntityRecognitionImage": {
      "instructions": {
        "focus": "Extract name, address, license number, and expiration date",
        "prioritize": "PERSON,ADDRESS,ID_NUMBER,DATE",
        "handleMultipleFormats": true,
        "context": "identity verification"
      }
    }
  }
}

Example - Receipt with Instructions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/receipt-photo.jpg"
    }
  ],
  "actions": {
    "namedEntityRecognitionImage": {
      "instructions": {
        "focus": "Extract vendor name, total amount, date, and item names",
        "prioritize": "ORGANIZATION,MONEY,DATE,PRODUCT",
        "handleMisalignment": true,
        "context": "expense report processing"
      }
    }
  }
}

Output:

{
  "id": "abc890",
  "results": [
    {
      "objectKey": "images/business-card.jpg",
      "namedEntityImage": {
        "isSuccess": true,
        "result": {
          "PERSON": ["John Doe"],
          "ORGANIZATION": ["Acme Corp"],
          "EMAIL": ["john@acme.com"],
          "PHONE": ["555-1234"]
        },
        "error": null
      }
    }
  ]
}

Complex Multi-Action Examples​

Example 1: Complete Document Analysis​

Process a document with multiple text actions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/quarterly-report.pdf"
    }
  ],
  "actions": {
    "textSummarization": {
      "maxWordCount": 200
    },
    "textClassification": {
      "classes": [
        "financial_report",
        "business_plan",
        "market_analysis",
        "internal_memo"
      ]
    },
    "textMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "report:type": "Quarterly|Annual|Monthly|Special",
          "report:period": "Q1|Q2|Q3|Q4|FY",
          "department:name": "Finance|Sales|Operations|Marketing",
          "has:financials": "yes|no",
          "keywords:tags": "revenue|expenses|growth|performance"
        }
      ]
    },
    "namedEntityRecognitionText": {},
    "textEmbeddings": {}
  }
}

Example 2: Comprehensive Image Processing​

Analyze product images with multiple actions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "images/product-photos/item-001.jpg"
    },
    {
      "path": "images/product-photos/item-002.jpg"
    }
  ],
  "actions": {
    "imageDescription": {
      "maxWordCount": 75
    },
    "imageClassification": {
      "classes": [
        "electronics",
        "clothing",
        "furniture",
        "food",
        "toys",
        "other"
      ]
    },
    "imageMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "product:category": "Electronics|Clothing|Home|Sports|Beauty",
          "product:condition": "New|Used|Refurbished",
          "image:quality": "High|Medium|Low",
          "background:type": "White|Studio|Lifestyle|Transparent",
          "has:watermark": "yes|no"
        }
      ]
    },
    "imageEmbeddings": {}
  }
}

Example 3: Mixed Document Types​

Process different document types with tailored actions:

{
  "version": "context.api/v2",
  "objectKeys": [
    {
      "path": "documents/contract.pdf"
    },
    {
      "path": "documents/invoice.pdf"
    },
    {
      "path": "documents/report.pdf"
    }
  ],
  "actions": {
    "textClassification": {
      "classes": [
        "contract",
        "invoice",
        "report",
        "letter",
        "form"
      ]
    },
    "textMetadataGeneration": {
      "kSimilarMetadata": [
        {
          "document:category": "Legal|Financial|Administrative|Technical",
          "urgency:level": "high|medium|low",
          "requires:signature": "yes|no",
          "keywords:tags": "important|review|action_required|reference"
        }
      ]
    },
    "namedEntityRecognitionText": {},
    "textEmbeddings": {}
  }
}

See Also​