Document intake automation

Turning faxed and scanned requisitions into structured orders before a sample reaches the bench. Scope still being confirmed.

Problem

Paper requisitions keyed by hand set the floor on turnaround time.

What I did

In flight: extraction with a human confirmation step rather than full automation.

Result

To be measured. No numbers claimed until they can be sourced.

Role

Product manager and builder

Timeframe

In progress

Team

To be staffed

Stack

Python, document AI, LIS API

TBD

keying time removed

TBD

requisitions per day

Context

Placeholder. Orders still arrive on paper from a long tail of small practices, and the intake queue sets the floor on turnaround time for everything behind it.

The problem

Placeholder. Two staff key requisitions by hand each morning; errors are caught late, at the bench, when the sample is already in process.

Why this and not something else

Placeholder. The strongest argument here is likely against pushing every small practice onto an electronic order form, which shifts the cost onto the client and does not remove the paper.

Decisions and tradeoffs

Placeholder. Expect a human-in-the-loop confirmation step in v1 rather than full automation, with confidence thresholds tuned to favour review over silent error.

Intake queue. Screenshots to be captured once v1 is in a reviewable state.

Constraints

Placeholder. PHI on every page, no third-party document service without review, and no change to the bench workflow.

Outcome

Placeholder. No outcome to report yet; this page exists to hold the structure while the work is in flight.

TBD

Manual keying time removed per morning

TBD

Requisitions per day through the queue

TBD

Review rate at the chosen confidence threshold

Placeholder paragraph for the outcome narrative once numbers exist and can be sourced.

What I'd do differently

Placeholder. To be written honestly after the work ships rather than invented now.

Open to technical product management roles.

ykrsandson@gmail.comLinkedIn