On-Prem · v4.0 Enterprise

Financial document intelligence, fully self-hosted.

Ingest thousands of pages, parse complex financial layouts, and chat with your knowledge base — running entirely on your own infrastructure.

$ bun run dev — frontend · api · workers · socket · monitoring

AI Ingestion Pipeline
Upload
Queue
Page Split
Layout Parse
Markdown
Chunk
Embed
Vector Store
RAG Ready
>95%
OCR accuracy
10,000
pages / file
<1s
retrieval latency
>100
pages / min

Built for high-volume financial documents

Every layer — parsing, embedding, retrieval — engineered for complex, tabular, scanned financial reports.

Layout-aware OCR

Qwen3-VL-8B reads scanned and mixed-layout financial PDFs with >95% OCR accuracy.

Financial table understanding

Nested, merged-cell and multi-page tables reconstructed with row continuity intact.

pgvector retrieval

Semantic search over millions of chunks with sub-second retrieval latency.

Conversational RAG

Streaming answers with inline source citations across files and workspaces.

Enterprise RBAC

Super-admin, admin and user roles with realtime audit trails and governance.

Fully on-prem

Air-gapped deployable. No mandatory cloud. One-command Docker Compose ops.

Financial intelligence

Understands the documents finance teams actually use

From audited annual reports to scanned invoices — including rotated tables, merged cells, and statements that span dozens of pages.

Balance sheetIncome statementCash flowBank statementInvoiceAccounting ledgerAudit reportAnnual reportFootnotes & appendixRotated tables
>90%
Table extraction
>90%
Layout reconstruction
>95%
Row continuity
Per-tenant
Workspace isolation

Deploy your private RAG platform today

One command. Your infrastructure. Your data never leaves the building.