Product
From document to structured data, checked along the way.
Jannat AI reads every logistics document, compares the documents on a shipment against each other, keeps everything searchable and reportable, and pushes structured data into the system you already use. Four parts of one workflow, explained below.
How it works
Upload anything. Get structured fields back.
Upload any PDF, JPG or PNG from any carrier, vendor or shipper. No templates, no configuration, no IT project.
Step 1
Upload
Drag and drop, bulk upload, or send by API. Scanned or native, any vendor, any carrier, any format.
Step 2
Classify
The document type is identified automatically. Nothing to label or sort first.
Step 3
Extract
Every relevant field is located, read and normalized: line items, totals, dates, addresses.
Step 4
Export
Download CSV or Excel, push by webhook, or pull through the API.
Documents it reads
Invoices
Carrier invoices, vendor invoices and freight bills: vendor, invoice number and date, due date, line items, total, tax, PO number.
Bills of lading
Standard, straight or order: shipper, consignee, carrier, PRO number, origin, destination, commodity, weight, pieces.
Delivery notes and PODs
Delivery date, recipient, address, items and quantities, signature status, order number. Works from phone photos.
Packing lists
Supplier, PO number, SKUs, quantities, carton count, net and gross weight, container information.
Commercial invoices
Exporter, importer, HS codes, country of origin, currency, total value, terms.
Fuel and expense receipts
Date, merchant, amount, fuel type, gallons, location, driver. Photos included.
Works with whatever you receive
Logistics teams do not control how their vendors format documents. Jannat AI does not ask them to.
- Native PDFs from carrier portals and TMS systems
- Scanned PDFs, including low-resolution fax scans
- JPG and PNG photos taken on a phone in the field
- Multi-page PDFs with multiple documents inside
- Any vendor or carrier layout, with no template library
- Handwritten fields alongside printed text
Built around your workflow
- Bulk upload: send a whole week of documents at once.
- API ingest: send documents from your TMS, portal or internal tools.
- Webhook notifications: get a call the moment a document finishes.
- Human review: low-confidence fields are flagged before export.
- Confidence scoring: every field carries a score, so uncertain reads are never silent.
- Normalized output: ISO dates, numeric currencies, structured addresses.
- Team access and retention controls: role-based access, with storage time set per workspace.
Document validation
Catch the mismatch before CBP does.
Jannat AI reads every document on a shipment, compares them against each other, and flags the fields that disagree, before you file.
Everything matches except the one thing that doesn't
A commercial invoice says a shipment weighs 1,240 kilos. The bill of lading on the same container says 1,420. Consignee matches. Container number matches. HS code matches. Carton count matches.
So nobody catches it. It gets filed.
That is not carelessness. It is four in the afternoon with nine more entries in the queue, and the error is invisible unless someone reads two documents side by side, field by field, every time.
Upload the whole shipment
Drop in the bill of lading, the commercial invoice, the packing list. Any format: crooked scans, stamped faxes, phone photos.
Every field extracted
Consignee, HS code, container and seal numbers, gross and net weight, carton count, port of loading.
Disagreements flagged
Fields that should match across documents and don't are flagged before you export anything. You see both values side by side and decide.
What gets compared
| Field | Compared across | Why it matters |
|---|---|---|
| Gross and net weight | Invoice, BOL, packing list | Weight discrepancies trigger holds and VGM issues |
| HS code | Invoice, entry data | Wrong classification means wrong duty |
| Consignee and shipper | All documents | A mismatch stalls release |
| Container and seal number | BOL, packing list | A transposed digit loses the container |
| Carton and piece count | Invoice, packing list, BOL | Short-ship and over-ship disputes |
| Value and currency | Invoice, entry data | Undervaluation is a penalty exposure |
| Country of origin | Invoice, certificate of origin | Marking and trade-remedy exposure |
"Our documents are too messy for that"
Most tools work on clean PDFs and fall apart on a faxed copy with a stamp across the corner and handwriting in the margin. Ours was built for the second kind, because that is what actually arrives.
No templates. No field mapping. No telling it where to look on the page.
Send us the ugliest document you have this week. If it cannot read it, we will tell you straight.
Dashboard and analytics
Every document you process stays findable.
Each upload and extraction is saved to a searchable history. Find any document by consignee, container, entry number or date, without opening PDFs one by one, then report across all of it.
Search everything
Search across all extracted data and uploaded documents. Look up a shipment by consignee, container, entry or date and see every document and extraction tied to it.
Volume reporting
See how many documents you process, by type and by month, so seasonal patterns and sudden changes are visible. Export reports as CSV or Excel.
Vendor and counterparty view
See documents grouped by the carrier, vendor, shipper or consignee on them, and how each one shows up across your history.
Discrepancy reporting
See which counterparties produce documents that disagree with others on the shipment, and which fields. Repeated mismatches from the same source are a pattern worth raising with them, instead of fixing quietly each time.
Why counterparty discrepancies matter
A single mismatch is a correction. The same mismatch from the same supplier or carrier every month is a process problem upstream. Discrepancy reporting turns one-off catches into evidence you can take back to the counterparty.
Teams also get processing status for every document in the queue (pending, processing, done or flagged for review) and a record of who uploaded and exported what.
Integrations
Your data lands where you already work.
45+ built connections across TMS, ERP, accounting and freight platforms. No middleware, no re-keying, no IT project.
Extraction that stops at a download isn't finished
Pulling fields off a document is half the job. If somebody then copies those fields into your TMS by hand, you have moved the typing rather than removed it.
Jannat AI pushes structured data into the system your team already works in, so the document is handled end to end.
By category
TMS, transportation management
ERP
Accounting and finance
Freight and customs
No connector needed
CSV export
One click. Opens in anything.
Excel export
Formatted, ready to work in.
Webhooks
Push extractions to your system the moment they finish.
REST API
Full programmatic access, with documentation.
We'll confirm before you commit
Tell us which system you file into and we will confirm exactly what the connection does, what it writes, and what setup takes, before you buy, not after. If it is not one of the 45, the API and webhooks cover it.
See who it is built for
The same workflow, applied four ways.