Platform · Data sources
The fewer copies, the better the diligence.
We connect directly to your raw ledgers, filings, and payment systems — and only ask for documents when we have to. Every number traces to its source.
Data architecture
L0 first. Always.
Our four-layer data model — connectors, normalization, computation, and attribution — keeps a stable link from every output back to a raw record.
Trusted data
We connect to source. Not summaries.
Every metric is rebuilt from raw records — filings, ledgers, market data, court dockets — across US and India jurisdictions.
US filings & courts
India regulators
Accounting & banking
Market & alternative data
How it works
Three paths to source data.
Direct connectors
OAuth-based pulls from accounting, banking, and ERP systems. We never see your credentials.
- QuickBooks Online, Xero, NetSuite — full GL extraction
- Stripe, Plaid — transaction-level banking data
- Tokens stored in GCP Secret Manager, referenced by pointer
- Read-only scopes wherever the API supports them
Public data, fetched live
Filings and regulatory data pulled fresh on each analysis. No stale snapshots.
- SEC EDGAR — 10-K, 10-Q, 8-K, S-1, DEF 14A
- MCA21 (India) — 2.8M+ company filings
- NSE / BSE — financial results, shareholding, insider trades
- CourtListener, USPTO, Indian Kanoon — litigation and IP
Documents you upload
When source connectors aren't an option, we extract from documents — and tell you what level of confidence to attach.
- PDF, XLSX, DOCX, CSV — virus-scanned, max 50MB
- Stored in GCS with private ACL, signed-URL access
- Source attribution preserved on every extracted figure
- Confidence score per extraction (HIGH / MEDIUM / LOW)