Inside Logiintra: how one document becomes a live operation

At 9:42 am, a shipping line emails a booking confirmation to a forwarder's operations inbox. At 9:43, a job exists in the ERP — customer linked, containers recorded, cutoffs on the calendar, compliance queued. Nobody typed anything. Here's what happened in that minute inside Logiintra.

Step 1 — Capture: the inbox is the front door

Logiintra watches the operations mailbox. Every attachment — PDF, scan, even a photo of a document — is pulled in, fingerprinted (so duplicates and re-sends are recognised instantly) and queued for reading. Nothing depends on someone remembering to upload a file.

Step 2 — Read: OCR turns pixels into text

The document image becomes machine-readable text, whatever the layout — a clean carrier PDF or a skewed scan from a CFS office. This is table stakes; the interesting part is what happens next.

Step 3 — Understand: fields, not just words

The AI layer classifies the document (booking confirmation? BL? vendor invoice?) and extracts its meaning: which string is the booking number, which party is the shipper, which dates are the gate and SI cutoffs, which amounts are freight versus charges versus tax. Extraction runs at roughly 95% field accuracy, and every field carries a confidence score.

Step 4 — Match: connect it to your business

Data without context is noise. Logiintra matches the document to your records — the customer, the open quotation, the existing job if the document is a revision. A booking amendment updates the job it belongs to; it doesn't create a duplicate. This matching step is where most "automation" tools stop and ours begins.

Step 5 — Act: the operation runs

Depending on the document type, the system creates or updates the job, schedules the milestones, prepares e-Way Bills and GST entries, and posts matched vendor invoices to accounts. The books don't wait for month-end typing — they follow the operation in real time.

Guardrails, not blind faith: anything below the confidence threshold — an unusual charge, a mismatched total, an unknown party — lands in a review queue for a human decision. The AI does the typing; your team keeps the judgement.

Why this architecture matters

Because the AI is inside the ERP rather than a separate tool, there's no integration to build, no sync to break, and no second system to log into. That's what makes it affordable at SMB scale — and it's why 300+ companies now run lakhs of transactions a month through this pipeline.