What is RecordEngine?
RecordEngine is an AI-powered document processing platform that runs entirely on your own hardware — inside your office, your data centre, or your private cloud server. It reads any document you throw at it (invoices, contracts, receipts, meeting notes, audio recordings), extracts the structured data automatically using a locally-hosted AI model, and feeds that data directly into your existing systems via webhooks and a REST API. No document ever leaves your network. No cloud subscription. No per-page fees.RecordEngine’s tagline is “Cut the Cloud. Keep the Intelligence.” — the AI runs locally on your GPU, not on a third-party server.
How it works
A document enters RecordEngine
Via web upload, a watched hot folder, or email attachment — RecordEngine detects a new file and queues it for processing.
The AI reads and extracts
A locally-hosted multimodal AI model reads the document — whether it’s a clean PDF, a photo of a receipt, a scanned contract, or an audio recording. It extracts every relevant field, generates a plain-language summary, and assigns a confidence score.
You review and approve
The extracted data appears in a clean three-panel interface. You verify the fields, correct anything the AI missed, and approve the document. Documents with low confidence scores are automatically flagged for priority review.
Your systems get the data
The moment a document is approved, RecordEngine fires an outbound webhook with the full structured payload — vendor, date, amounts, line items, confidence score, and any CRM correlation IDs. Your ERP, CRM, or accounting system receives clean structured data without any manual re-entry.
What makes it different
Fully Offline
Every AI inference call goes to a model running on your own GPU. Nothing touches the internet after installation. Designed for environments with strict data residency requirements.
No Per-Document Pricing
One hardware purchase, one annual software licence. Process 10 documents a day or 10,000 — the cost is the same. No token fees, no page fees, no seat fees.
Model-Agnostic
The AI model is swappable. RecordEngine ships with a best-in-class open-source model, but you can substitute any Ollama-compatible model without touching the platform code.
Integration-First
Every document, field, and event is accessible via REST API and outbound webhooks. Bidirectional CRM correlation is built in — not bolted on.
Supported file types
RecordEngine processes all of the following natively — no plug-ins or additional software required:| Category | Formats |
|---|---|
| Documents | PDF, DOCX, TXT, HTML |
| Images | JPG, JPEG, PNG |
| Spreadsheets | XLSX, CSV |
| Audio | MP3, WAV, M4A |
Built for data-sensitive environments
RecordEngine is used by organisations where document data is too sensitive for cloud processing:Accounting & Financial Services
Accounting & Financial Services
Process client invoices, receipts, and financial statements without sending financial data to third-party cloud APIs. Structured output flows directly to QuickBooks, Xero, or your own ERP via webhook.
Law Firms
Law Firms
Contract review, court document processing, and client file management — all processed on your own infrastructure, with a full audit log of every action taken on every document.
Multinational Corporations in China
Multinational Corporations in China
RecordEngine is fully compliant with China’s PIPL, DSL, and CSL data residency requirements. Documents processed in China stay in China. Built-in support for Chinese Fapiao (发票) and Business License (营业执照) extraction.
Healthcare
Healthcare
Patient records, insurance claims, and referral letters processed locally — no PHI leaving your network.
Government & Defence
Government & Defence
Air-gap installation available. Deploy from a USB bundle with no internet connection required at any point during installation or operation.
Platform at a glance
| Version | 2.5.5 |
| Extraction profiles | 7 built-in + unlimited custom |
| Document statuses | 9 (New → Processing → Needs Review → Ready → Approved → Export → Exception → Missing Info → Archived) |
| API | REST API + outbound webhooks |
| AI model | Locally-hosted, open-source multimodal model |
| Supported languages | English, Chinese (Simplified), and any language supported by the underlying model |
| Deployment | Docker (Linux or Windows), private cloud, or air-gap USB bundle |
| Authentication | Multi-user, role-based (Admin / User) |