File Integration

Upload and process documents to extract memories

POST /integrations/files

cURL

curl -X POST 'https://api.memchat.io/integrations/files' \
  -H 'X-API-Key: YOUR_API_KEY' \
  -F 'file=@document.pdf' \
  -F 'category=learning' \
  -F 'extract_mode=comprehensive'

import requests

url = "https://api.memchat.io/integrations/files"
headers = {
    "X-API-Key": "YOUR_API_KEY"
}

files = {
    'file': open('document.pdf', 'rb')
}

data = {
    'category': 'learning',
    'extract_mode': 'comprehensive'
}

response = requests.post(url, headers=headers, files=files, data=data)
result = response.json()

const formData = new FormData();
formData.append('file', fileInput.files[0]);
formData.append('category', 'learning');
formData.append('extract_mode', 'comprehensive');

const response = await fetch('https://api.memchat.io/integrations/files', {
  method: 'POST',
  headers: {
    'X-API-Key': 'YOUR_API_KEY',
  },
  body: formData,
});

const result = await response.json();

Response

{
  "integration_id": "file_int_456def",
  "filename": "research_paper.pdf",
  "file_size": 2458697,
  "status": "completed",
  "extraction_stats": {
    "total_pages": 15,
    "text_length": 45678,
    "chunks_created": 23,
    "memories_extracted": 12,
    "categories_detected": ["learning", "career", "interests"]
  },
  "memories": [
    {
      "memory_id": "mem_789ghi",
      "content": "The study demonstrates that machine learning algorithms can improve medical diagnosis accuracy by 23% when trained on diverse datasets.",
      "category": "learning",
      "importance": 0.87,
      "page_number": 3
    }
  ],
  "created_at": "2024-01-15T16:30:00Z",
  "completed_at": "2024-01-15T16:32:15Z"
}

Upload documents to automatically extract and store memories from their content.

cURL

curl -X POST 'https://api.memchat.io/integrations/files' \
  -H 'X-API-Key: YOUR_API_KEY' \
  -F 'file=@document.pdf' \
  -F 'category=learning' \
  -F 'extract_mode=comprehensive'

import requests

url = "https://api.memchat.io/integrations/files"
headers = {
    "X-API-Key": "YOUR_API_KEY"
}

files = {
    'file': open('document.pdf', 'rb')
}

data = {
    'category': 'learning',
    'extract_mode': 'comprehensive'
}

response = requests.post(url, headers=headers, files=files, data=data)
result = response.json()

const formData = new FormData();
formData.append('file', fileInput.files[0]);
formData.append('category', 'learning');
formData.append('extract_mode', 'comprehensive');

const response = await fetch('https://api.memchat.io/integrations/files', {
  method: 'POST',
  headers: {
    'X-API-Key': 'YOUR_API_KEY',
  },
  body: formData,
});

const result = await response.json();

Request Parameters

file · file · required

Document file to process. Supported formats: PDF, DOCX, TXT, MD, HTML

category · string

Default category for extracted memories. One of: identity, career, interests, relationships, health, finance, learning, travel, productivity, private

extract_mode · string · default: "smart"

Extraction strategy. Options: comprehensive, smart, highlights, summary

chunk_size · number · default: "1000"

Maximum size of text chunks for processing (characters)

overlap · number · default: "200"

Overlap between chunks to maintain context

auto_categorize · boolean · default: "true"

Whether to automatically categorize extracted memories

metadata · object

Additional metadata for the document

Show Metadata Properties

title · string

Document title

author · string

Document author

source · string

Source or origin of the document

tags · array

Custom tags for the document

date_created · string

When the document was created

Response

integration_id · string

Unique identifier for the file integration

filename · string

Name of the uploaded file

file_size · number

Size of the uploaded file in bytes

status · string

Processing status: uploaded, processing, completed, error

extraction_stats · object

Statistics about the extraction process

Show Extraction Statistics

total_pages · number

Number of pages in the document

text_length · number

Total characters extracted

chunks_created · number

Number of text chunks created

memories_extracted · number

Number of memories extracted

categories_detected · array

Categories detected in the content

memories · array

List of extracted memories (if processing is complete)

Show Memory Properties

memory_id · string

Unique identifier for the extracted memory

content · string

Extracted text content

category · string

Assigned category

importance · number

Calculated importance score

page_number · number

Source page number in document

created_at · string

ISO 8601 timestamp when upload was initiated

completed_at · string

ISO 8601 timestamp when processing completed

Response

{
  "integration_id": "file_int_456def",
  "filename": "research_paper.pdf",
  "file_size": 2458697,
  "status": "completed",
  "extraction_stats": {
    "total_pages": 15,
    "text_length": 45678,
    "chunks_created": 23,
    "memories_extracted": 12,
    "categories_detected": ["learning", "career", "interests"]
  },
  "memories": [
    {
      "memory_id": "mem_789ghi",
      "content": "The study demonstrates that machine learning algorithms can improve medical diagnosis accuracy by 23% when trained on diverse datasets.",
      "category": "learning",
      "importance": 0.87,
      "page_number": 3
    }
  ],
  "created_at": "2024-01-15T16:30:00Z",
  "completed_at": "2024-01-15T16:32:15Z"
}

Extraction Modes

Comprehensive

Extracts memories from all significant text passages, creating detailed memories for each section.

Smart

Uses AI to identify the most important passages and concepts, creating focused memories.

Highlights

Extracts only key insights, findings, and actionable information.

Summary

Creates a single comprehensive memory summarizing the entire document.

Supported File Types

PDF

• Text extraction from standard PDFs
• OCR for scanned documents
• Preserves page structure and formatting

Word Documents (DOCX)

• Full text extraction
• Maintains document structure
• Extracts embedded content

Text Files (TXT, MD)

• Direct text processing
• Markdown formatting recognition
• Fast processing

Web Pages (HTML)

• Content extraction from HTML
• Removes navigation and ads
• Focuses on main content

Error Responses

400 · object

Invalid file or parameters

{
  "error": "Unsupported file type",
  "code": "INVALID_FILE_TYPE",
  "supported_types": ["pdf", "docx", "txt", "md", "html"]
}

413 · object

File too large

{
  "error": "File size exceeds limit",
  "code": "FILE_TOO_LARGE",
  "max_size_mb": 50,
  "file_size_mb": 75
}