Proto Company 07 — shared for feedback · leave a note under any section · v11 · 0 notes
A Peterson & Gill proto company — shared with you for feedback

Efficient healthcare delivery, on one connected architecture.

Proto Company 07 is a processing layer for health programmes: any stream of field data in — dashboards, decisions and daily work out. It connects screening to the end of the care pathway and enables every health worker along the way. Below, the idea is rendered on TB — where our market-shaping expert has convinced grant providers and ministries to get the screening cost down to 1¢ per person through pay-per-use licenses.

The architecture

A connector from screening to the end of the pathway — enabling every health worker along the way.

Diagnostic and screening inputs flow in from the field — X-rays, forms, lab results, stock feeds. One engine interprets them: triage, forecasts, effort optimisation. Administrators get the dashboard view — a real-time picture of stock-outs and hotspots. Community health workers get guided pathways in the chat they already use — trained, incentivised, and recording back so the system stays current. Swap the configuration and the same engine serves another ministry.

Present this ⤢
Screening inputs
diagnostics from the field
Portable X-ray AI read of each film
Screening forms symptoms & surveys
Lab results confirmation tests
Pharmacy stock feeds consumption & levels
… anything with an adapter
Processing engine
interprets every stream
Triage & risk scoring who needs attention, how urgently
Forecasts stock-outs & hotspots, before they happen
Effort optimisation who goes where tomorrow
Dashboards
a real-time picture of the field
Ministry / admin stock-outs, hotspots, where to send effort next
Supervisors alert & triage — alerts arrive on WhatsApp too, with a dashboard login on top; they assign the case to the ASHA
live · de-identified · auditable
CHW pathways
start on the supervisor's assignment — in the chat they already use
Guided tasks the assigned visit, walked through on WhatsApp
Train & refresh micro-modules when the data says they're needed
Incentives verified actions, rewarded
Report-back one tap — the system updates itself
CHW report-back flows straight back into the processing engine
Program packs — plug-and-play verticals
TB screening — live, this site Mental health — incl. stock-out response Agriculture / other + next vertical

Same engine underneath. A vertical is a configuration file — screening rules, patient sequence, dashboards, incentives — not a rebuild. What you see below is the TB pack running.

Rendered on TB — how it works

One scan sets off an automated cascade — all the way to treatment.

A chest X-ray on an existing machine is read by AI in seconds. The instant it flags TB, the system takes over the hand-offs that usually leak — alerting a health worker, booking confirmation, enrolling treatment, screening the household, and logging everything to the national register.

Present this ⤢
1¢ a scan!

At the moment, most grant & country efforts go toward procuring technology that gives deterministic pathways like more efficient ways to identify positive patients.
We're proposing far more efficient costing models, so the focus moves to where it should be: eradicating TB, treating one patient at a time.

On the ground

What the community health worker sees.

The moment a scan is positive, a notification reaches the health worker who covers that household — on WhatsApp, the tool they already use every day. No new app, and no chatbot: just standard WhatsApp Business template messages with tap-to-reply buttons. The worker taps; the system files the paperwork. Every branch is a pre-approved protocol — including the hard moments, like a family hesitant over stigma, which triggers the standard counselling audio and a supervisor notification.

Present this ⤢
TB
TB SentinelWhatsApp Business · automated

Why this matters: the cascade normally leaks at every hand-off — referral lost, patient never tested, treatment never started. Here the referral, lab order, result sync, treatment enrolment, register entry and household scheduling all file automatically off the worker's button taps. The health worker does the human part — showing up, building trust, dispelling fear — and never touches a form.

Deliberately boring technology: everything shown is standard WhatsApp Business — template messages, quick-reply buttons, media messages. Pre-approved protocol branches, not an AI chat. That's what makes it deployable inside a ministry's existing approval processes, and usable by workers with no training curve.

For the programme

What the ministry sees.

Every scan, confirmation and treatment start writes to the national register in real time. Administrators get a live picture of coverage, caseload and cascade completion — and, crucially, where to send the next health worker or mobile X-ray unit.

Present this ⤢
National TB dashboardLive · this month
People screened
0
64% on idle machines
CAD-flagged
0
presumptive TB
Confirmed
0
bacteriological
On treatment
0
auto-onboarded
Cascade complete
0
confirmed → treated
Avg. screen→treat
0
was 3–5 weeks
Caseload by district — where effort is needed
Ganeshpur▲ needs attention1.4% pos
Rampur0.8% pos
Kotwali0.6% pos
Sadar0.5% pos
The cascade, in numbers
Screened · ~1¢ per patient48,200
CAD-flagged10,600
Confirmed338
On treatment297
All 48,200 screens ≈ $482. The cents live at the top of the cascade — the savings appear below it.
Auto-generated recommendation Ganeshpur is running 2× the state-average positivity with only 62% CHW coverage and a 71%-idle X-ray unit at the district hospital. Suggested action: deploy 4 additional CHWs and activate 1 mobile X-ray unit here next cycle.
The cherry on top

Why it costs about a cent.

The status quo buys hardware and per-machine software licenses. We do neither. We read on the idle base a country already owns, and charge a cloud-or-offline license priced per use — so the marginal cost of screening one more person approaches a penny. The play: reduce the cost of what the programme already spends on, redirect that funding down the cascade, and show the results.

Present the calculator ⤢
01 · Reduce
Screening cost collapses to ~1¢ per patient, on machines the country already owns.
02 · Redirect
Freed budget moves to where the programme currently overspends — confirmation, follow-up, treatment starts.
03 · Results
More people on treatment per dollar — visible live on the ministry dashboard.
The model

Existing infrastructure × cloud-offline license = ~1¢ per read, pay-per-use.

No fleet to buy. No per-machine license to renew. The AI runs in the cloud where there's connectivity and on a small offline edge box where there isn't — syncing when the link returns — so remote, high-burden districts aren't left out. Countries pay only for scans actually read.

Infrastructure
Reuses the digital X-ray machines already installed and idle
Licensing
Cloud or offline edge — priced per scan, not per device
Commercials
Pay-per-use · no capex · scales down as easily as up
Teleradiology
Optional upgrade — never bundled into the 1¢ read
Per-machine TB software license
$2,850
Recurring, per machine — the status-quo cost we remove.
Ultraportable X-ray, bought new
$31,000
A fleet purchase. We read on the idle base instead.
Our marginal cost per AI read
~$0.01
On a pre-existing digital X-ray, marginal cost approaches zero.

What it looks like at scale

Because onboarding is automated, cases found and people started on treatment scale together. Pick a site and slide the population screened.
Population screened250,000
Cost per case started on treatment
Screening cost
~$0.01 read + confirmation of flags
Confirmed cases
Auto-onboarded to treatment
Cost per case vs current active case-finding
Our system (incl. treatment-start)
Country ACF benchmark
What the pilot must prove

Everything ladders to three claims.

01
Utilization unlock is real
A large share of reads happen on machines the country already owns but wasn't using for TB. If true, our cost structure is structurally lower than buying fleets.
Target ≥60% of reads on pre-existing, idle machines
02
The cascade holds
Positives tracked screen → confirm → treatment start, leakage measured at every step, written to the national register. Automated hand-offs are how it holds at scale.
Measure leakage at every step · automate what usually leaks
03
Unit cost beats the status quo
Cost per case started on treatment below the country's current active-case-finding cost — a stricter bar than "per confirmed case".
Beat local ACF · ref: Nigeria CXR+AI ≈ $635–1,198 / confirmed case