Document data you can trust.
CipherCloud reads invoices, bills of lading, KYC forms and agreements, returns clean structured fields, and scores every one for confidence — so your team only reviews what needs reviewing.
- 1,000 free tokens
- No credit card required
- Isolated, encrypted workspaces
Built for document-heavy, regulated work
- BankingKYC, statements, trade finance
- PropertySPAs, agreements, title docs
- LogisticsInvoices, delivery orders
- FreightBills of lading, customs forms
- InsuranceClaims, policies, schedules
- ManufacturingPOs, packing lists, specs
- Finance opsAP invoices, receipts
- GovernmentPermits, filings, IDs
- typical end-to-end processing
- <30s
- file formats accepted
- 5
- free tokens to start
- 1,000
- per extracted field
- RM 0.01
One audited path, from upload to your system
Every document follows the same four stages. Confident results flow straight through; uncertain ones stop for a person.
Ingest
Upload in the app, forward by email, or send through the API. PDF and image scans, batches welcome.
Read
OCR captures the text and the table structure of every page — line items keep their rows and columns.
Extract
A language model maps the page to your document type’s schema and returns named, typed fields.
Score
Every field gets a confidence score. Your workspace threshold decides what happens next.
Results are available as JSON or CSV, by webhook, API or export — no one has to touch them.
The document waits in a review queue with low-confidence fields flagged; tokens are charged only once it completes.
Everything a document pipeline needs
The building blocks of a production extraction workflow — without the integration tax.
Template-free extraction
OCR feeds a language model that returns clean, labelled, schema-ready fields — including full line-item tables. No brittle zones or regex to maintain.
Batch ingest
Drag and drop, email in, or pipe through the API by the hundred. PDF, JPEG, PNG, TIFF, BMP — up to 10 MB each.
Fast by default
Most documents resolve end to end in under 30 seconds.
Structured output
JSON for systems, CSV for spreadsheets, and an inline review view for everyone else.
API, SDK and MCP
A REST API, an OpenAI-compatible /v1 endpoint, and an MCP server your AI agents can call directly.
Pay as you go
Token pricing in MYR — one token per extracted field. No subscription, no minimum.
Three ways in. One line of setup.
Wire extraction into an existing system, reuse the OpenAI SDK you already have, or hand it to your AI agents.
REST API
Sync or async extraction, batch upload, classification — authenticated with a workspace API key.
OpenAI-compatible
Point the official OpenAI SDKs at /v1 and change nothing else. Python, JavaScript, anything that speaks the API.
MCP server
20 tools for extraction, templates and review. Claude Code, Cursor and Windsurf call Cipher as a native tool.
curl -X POST https://cipher.springtech.ai/api/v1/public/extract/sync \
-H "X-API-Key: $CIPHER_API_KEY" \
-F "file=@invoice.pdf"Accuracy you can see. Data you control.
Regulated teams need more than fast extraction. They need to know which values to trust, who touched them, and where the documents live.
Confidence on every field
Each extracted value ships with a confidence score, so you know exactly what to trust and what to double-check.
Human review, built in
Documents below your workspace threshold are routed to review automatically — low-confidence values never slip through silently.
Workspace isolation
Every document, key and result is scoped to a workspace. Nothing crosses between teams or tenants.
Documents live in object storage scoped to your workspace and encrypted at rest.
OCR and extraction run on GPU servers we operate. Any fallback model provider is chosen and controlled by us, never by chance.
Workspace members, scoped API keys and admin controls keep access with the people who need it.
Key actions are recorded so you can trace who did what, and when, across your workspace.
Remove documents and their extracted results on demand — no retention you didn’t ask for.
Simple, pay-as-you-go tokens
One token is one extracted field. Buy a package when you need it — no subscription, no minimum. Every account starts with 1,000 free tokens.
Enterprise
20% offMaximum scale at lowest per-token rate
Get started- REST, OpenAI-compatible & MCP access
- Confidence scores + review queue
- Top up anytime via Stripe
Processing millions of pages? Talk to us about volume pricing
Frequently asked
Something else? Email the team.
Put your documents to work.
Start with 1,000 free tokens — no credit card, no commitment. Or talk to us about a pilot on your own documents.