Files can be uploaded to your Space through the web interface or programmatically via the API. Once uploaded, files are processed automatically and become available for search, AI interaction, and sharing.
Via the UI
Upload files through the Space interface using drag-and-drop or the file picker:
- Upload Files — select individual files from your device
- Upload Folder — upload an entire folder and its contents
- Create a Folder — organize files into folders before or after uploading
Multiple files can be uploaded simultaneously. Upload progress is shown in real time.
After upload, files are indexed for search and become available for AI interaction. Most documents and images process within seconds. Large media files take longer.
Via the API
Upload a file programmatically using a multipart form request:
import requests
with open("report.pdf", "rb") as f:
response = requests.post(
"https://api.zarkai.xyz/v1/storage/files",
headers={"x-api-key": "your_api_key"},
files={"file": ("report.pdf", f, "application/pdf")},
)
data = response.json()
print(data["fileId"]) # e.g. "file-55f3fa40-9883-45d5-a877-a68264d58929"
Response
{
"fileId": "file-55f3fa40-9883-45d5-a877-a68264d58929"
}
fileId is used to reference this file in subsequent API calls — chat requests, file management, and deletion.
Using Uploaded Files in Chat
Once a file is uploaded, reference it in chat requests via the file_ids parameter:
{
"messages": [
{"role": "user", "content": "Summarize this document"}
],
"workspace_id": "wks-xxxxx",
"file_ids": ["file-55f3fa40-9883-45d5-a877-a68264d58929"],
"tool_choice": "auto"
}
Processing by File Type
Different file types are processed differently after upload:
| File type | Processing |
|---|
| Documents (PDF, Word, etc.) | Text and structure extracted. Tables preserved as queryable data. Scanned documents processed with OCR |
| Images (PNG, JPG, etc.) | Visual content analyzed. Text extracted via OCR. Objects, scenes, and context identified |
| Data files (CSV, Excel, etc.) | Structure parsed, columns detected, data types inferred, formatting cleaned. Becomes a queryable table |
| Audio (MP3, WAV, etc.) | Available for AI questions immediately. Full transcription via Media Processor |
| Video (MP4, MOV, etc.) | Available for AI questions immediately. Full transcription and visual analysis via Media Processor |
| Code (Python, JS, etc.) | Content indexed for search and AI interaction |
| Category | Formats |
|---|
| Documents | PDF, Word (.docx, .doc), PowerPoint (.pptx, .ppt), Markdown, HTML, plain text, RTF |
| Images | PNG, JPG, JPEG, SVG, GIF, WebP, BMP, TIFF, HEIC |
| Videos | MP4, MOV, AVI, MKV, WebM |
| Audio | MP3, WAV, M4A, FLAC, OGG |
| Data | CSV, TSV, Excel (.xlsx, .xls), JSON, Parquet |
| Code | Python, JavaScript, and other common formats |
