The clinician's app
Electron, React, TypeScript, Node
Runs on Windows, Mac or Linux. Keeps working when the connection drops.
Our main system
the middle man
A.N.C.H.O.R.
Autonomous Navigator for Clinical Healthcare Orchestration & Reliability
ANCHOR keeps track of what a hospital has promised a patient, and chases the things that slip. It runs on the computer in front of the clinician and keeps working when the network does not.
What it is not
not an app
A post-discharge operating layer — not an app bolted on the side.
Not a patient app
Engagement alone does not close a referral that never booked.
Not concierge post-discharge
Boutique follow-up does not become hospital infrastructure.
Not a clinic with software
The product is the ledger and the orchestration, not a logo on a practice.
ANCHOR is the layer that writes each discharge promise down — owner, window, what counts as closed — and keeps chasing what is still open.
The parts
five parts
The parts
| Area | Part | What it does |
|---|---|---|
| What people see | H.E.L.M. | The screen a clinician works in, and the reasons behind what it suggests |
| How work moves | Autopilot bus | Passes events between parts, checking the rules each time |
| The helpers | Specialist agents | Each does one narrow job, and says so when it does not know |
| Working out what is urgent | OLRHM + TRIS | The maths that ranks what is closest to going wrong |
| The safety core | Shared kernel | The shared foundation everything else has to go through |
How work is ordered
priority not fear
Urgency you can argue with. A queue that is not sorted by fear alone.
An inspectable urgency score shows which clinical and social factors pushed a patient up — so a coordinator can disagree with a number instead of fighting a black box.
The worklist does not simply sort by who looks riskiest. It orders work by how much a minute of coordinator time is expected to reduce the chance a promise is missed — so a high-risk patient with nothing actionable today does not crowd out a patient whose medication loop closes with one call.
Ambient documentation tools write notes. Event feeds tell you something already happened. Human navigation programs put people on the phone. ANCHOR is the ledger those efforts run on: which promise is open, whose, and what closes it.
How it works
a person decides
How a piece of information travels through it
From an event arriving to a person deciding — every clinical call stays with a licensed clinician.
- 01
Take it in
Only with permission
Appointments, notes, device readings and workflow events arrive tagged with where they came from, who consented, how sensitive they are, and which rules apply.
- 02
Write it down
One record per promise
Every open promise becomes its own record: who owns it, when it is due, where it has got to, what has been tried, and whether we know the full picture.
- 03
Work out the pressure
Hazard, not a snapshot
The ranking maths recalculates how close a patient is to something being missed, every time a promise opens, is fixed, is missed, or is cancelled.
- 04
Put it in order
Show the reasoning
The to-do list puts the most useful next work first, and shows which open promises pushed it up the list.
- 05
A person decides
Always
A clinician makes the call. What was done, why, which version of the software suggested it, whether they overruled it, and what happened next are all recorded.
Where it runs
on your machine
It runs on the machine in front of you
ANCHOR is built to work where clinicians actually are, storing everything locally and encrypted. Shared services can sit beside it when a hospital needs them — nothing basic should depend on them.
Storage
SQLite with SQLCipher
Encrypted on the machine, with no setup. Patient names must never appear in plain text in the logs.
The thinking, done locally
Ollama, Whisper, deterministic rules
Transcription and AI run on the same machine. Anything an AI produces can never be the only reason care is stopped.
Shared services
Kafka, NATS, Vault, Terraform
For running across a whole hospital when the site is ready for shared infrastructure.
Next
live walkthrough
See it working.
We walk through the system live — the ledger, the boundary, and how a clinician stays in charge.