Healthcare & Clinics

Reports that stopped needing a chase

A diagnostics lab cut report turnaround from three days to four hours by fixing the handoffs, not the analysis.

This is an illustrative composite. It is drawn from delivery patterns across our work rather than published as a named client reference, and the figures describe the pattern rather than one audited engagement. We would rather label it than imply a reference we cannot put you in touch with.

Size
60 staff
Platforms
ForgeAtlas

Context

A 60-staff diagnostics lab serving around forty referring clinics. Sample processing was fast; getting the finished report back to the referring doctor was not. Three systems were involved and none of them talked: the lab instrument export, a shared drive, and email.

The problem

Median turnaround was three days, and most of it was waiting. A report sat on the shared drive until a pathologist noticed it, then sat in a drafts folder until someone remembered to send it. Referring clinics chased by phone, which cost the lab a person most of the afternoon and cost the clinics their patience.

What we did

  1. 01Mapped every handoff from instrument to inbox and timed each one. Analysis was 40 minutes of a 72-hour cycle.
  2. 02Replaced the shared drive with a pipeline: instrument export lands, is matched to its requisition, and appears in a pathologist's queue in priority order.
  3. 03Built the verification step as the only manual gate, with the context a pathologist needs on the same screen as the decision.
  4. 04Automated delivery to the referring clinic on verification, in the format each clinic had asked for.
  5. 05Put the referring clinic's own status page behind a link, which is what ended the chasing calls.

Architecture

Nothing here is a black box. You can read the architecture before you sign, and you own it after.

  1. 01IngestInstrument export watched and landed raw; the requisition match is a job, not a person.
  2. 02QueueA pathologist queue in Postgres ordered by clinical priority and age, with RLS scoping each pathologist to their own lab.
  3. 03VerificationJJ Atlas drafts the narrative section with citations to the result rows; the pathologist verifies and signs.
  4. 04DeliveryOn signature, delivery per referring clinic in their chosen format. Failure retries and escalates rather than sitting silent.
  5. 05StatusA scoped status page per referring clinic showing where each of their samples is.

Results

The numbers the work closed against.

target report turnaround, against a three-day baseline
4 hourstarget report turnaround, against a three-day baseline

Basis: Three days is the measured median. Four hours is the sum of the timed steps that remain when the shared-drive and drafts-folder waits are removed — analysis is 40 minutes of it.

design target from pod start to reports going out unchased
9 weeksdesign target from pod start to reports going out unchased

Basis: Scoped from the engagement shape: ingest pipeline, queue, verification surface and per-clinic delivery, sequenced so delivery lands last.

target reduction in inbound chasing calls
84%target reduction in inbound chasing calls

Basis: Assumes calls are driven by not knowing where a sample is, and that a scoped per-clinic status page answers that. Residual 16% covers clinical queries a status page cannot answer.

of the observed 72-hour cycle that is actual analysis
40 minof the observed 72-hour cycle that is actual analysis

Basis: A measurement, not a target: every handoff from instrument export to referring inbox was timed. It is stated because it is the reason the work is a handoff problem.

Stack

  • Next.js
  • Supabase Postgres
  • Edge Functions
  • pgvector
  • Vercel
  • Zod

Related work

85 staff

The waiting room emptied before the doctor arrived

Four clinics moved patient intake to the phone in the waiting room and got eleven hours a week back at reception.

11 hrs/week target reception time returned across four clinics

72 staff

A security review answered in an afternoon

An accountancy practice stopped losing three weeks per client to evidence-gathering by making the trail a by-product of the work.

4 hours target time to answer a 90-question security review

Does this look like your business?

Tell us what is slow and what it is costing. We will point you at the closest precedent we have and say plainly where yours would differ.

Scope it with us