Reference Documents

Upload

Tiga langkah: minta upload_url, lalu PUT bytes file langsung ke situ (tidak melewati server kami sama sekali), lalu konfirmasi selesai. Mendukung file hingga 50MB.

# 1. Minta presigned upload URL
curl -X POST https://api.platform.kakak.ai/v1/reference-documents \
  -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
  -d '{"name":"Rubrik Penilaian 2026","filename":"rubrik.pdf","mime_type":"application/pdf","size_bytes":4823000}'
# -> { "id": "ref_...", "upload_url": "https://...", "expires_at": "..." }

# 2. PUT file bytes langsung ke upload_url
curl -X PUT "$UPLOAD_URL" -H "Content-Type: application/pdf" --data-binary @rubrik.pdf

# 3. Konfirmasi upload selesai (memicu indexing)
curl -X POST https://api.platform.kakak.ai/v1/reference-documents/ref_.../complete \
  -H "Authorization: Bearer $KEY"

Lifecycle

Status: processing → ready (atau failed jika indexing gagal).

Di balik layar: file diparse, teks diekstrak, dan corpus retrieval dibangun otomatis. Status disinkronkan secara lazy saat GET /reference-documents/:id.

Supported Formats

PDF, DOCX, XLSX/XLS, PPTX, TXT, MD, CSV, JSON, XML. Maks 50MB per file. OCR belum tersedia di v1.

Delete

DELETE /reference-documents/:id hanya menghapus pointer — dokumen asli tetap ada di kakak.ai workspace. User bisa menghapus dokumen dari UI kakak.ai normal.