Discover
Watch official report, notice, download and sitemap pages.
The main intelligence engine
Every PDF becomes evidence-linked data.
SENNA discovers official MFI reports, preserves every file version, extracts financial tables, normalizes Nepali and English labels, validates the figures and turns verified metrics into comparable intelligence.
PDFDiscoverArchiveExtractValidateCompareDATA
Evidence pipeline
Not a PDF folder—an accountable processing chain
Every stage leaves a status, fingerprint and recoverable record. A replaced PDF creates a new version instead of destroying the old evidence.
Preserve
Download, hash and archive the original PDF in R2.
Read
Extract born-digital text and tables; OCR scanned pages when needed.
Normalize
Map inconsistent English/Nepali labels and units to standard metric codes.
Validate
Check ranges, balance-sheet relationships, identity and duplicate candidates.
Explain
Generate deterministic, source-linked trends only after evidence exists.
Official document catalogue
Reports, versions and extraction state
The original source remains visible beside the archived version and structured result.
Loading the report catalogue…
Period intelligence
Compare verified periods without reopening PDFs
Growth, deterioration and restatements are calculated from normalized values while preserving links to the source cells.
Choose an institution and two report period-end dates.
Report-derived intelligence
Peer Lens and Trend Lab
These views use only approved or corrected values. Pending machine extractions never enter rankings.
Peer Lens
Rank one verified metric across MFIs
Choose a metric and an exact period-end date.
Trend Lab
Follow one verified metric over time
Choose an institution and metric.
Restatement radar
Changes that require attention
Common language
Metric dictionary
One controlled definition can recognize many report labels.