Hospital network, 11 sites
A clinical knowledge assistant that never leaves the VPC
- Clinician research time
- -71%Clinician research time
- Patient data leaving the network
- 0 PHIPatient data leaving the network
Every shipment arrived as a bundle of scanned declarations, invoices and packing lists, in four scripts and no consistent layout. Eleven operations staff retyped them into the platform by hand. Handling time averaged just over six minutes per shipment, and the error rate rose predictably every afternoon.
Drishti behind a schema-first extraction contract, one schema per document type, with confidence gates tuned per field rather than per document. High-confidence fields post straight through. Anything below threshold lands in a review queue that shows the operator the crop of the original scan next to the extracted value, so a correction takes seconds rather than reopening the PDF.
Handling time per shipment fell to fourteen seconds median, with roughly 8% of fields still routed to human review — which the operations team considers the right number rather than a failure. The eleven staff were not laid off; the platform grew into the capacity instead. Volume is now 340,000 documents a month.
Hospital network, 11 sites
Challenger bank, 400k customers
Send the problem, not a spec. We'll come back within one business day with an honest read on whether we're the right team — and who to talk to if we're not.
attatvasystems@gmail.com · Replies within 1 business day, Kathmandu time (UTC+05:45)