# What we built (in plain English)

## The one-line answer
We built software that runs the day-to-day operations of our research site mostly on autopilot - finding and booking patients, handling the calls/texts/faxes/emails, keeping the schedule and records straight, chasing the money, and catching mistakes - so the team does far less manual busywork and fewer things slip through the cracks.

If you only remember one sentence: **most of the repetitive work a coordinator, recruiter, or finance person used to do by hand now happens automatically, with safety checks that catch mistakes - so the team's job shifts from doing the busywork to handling the exceptions the system flags.**

## Two dashboards, two audiences
- **Operations dashboard** - the team's daily cockpit. Leads, scheduling, patient messages, study progress. Where coordinators and recruiters live.
- **Finance dashboard** - the money side. What each study earns, what's owed, what's been paid, profit per study. Where finance and leadership live.

(Plus a public website where patients check eligibility and book, and a portal where referring doctors can send us patients.)

## What it actually does for us
Grouped by the job it does, not the tech:

**Getting patients in the door**
- Pulls leads from our website and ads automatically, and instantly checks which studies each person might qualify for.
- Tells recruiters who to call first - and tracks how fast we actually responded.
- Lets patients book pre-screens and visits online, then texts/emails reminders so fewer people no-show.
- Automatically wins back people who started a form but didn't finish; cancellation re-engagement is built and runs in observe mode (sends gated off).

**Running the day**
- One screen shows every visit: who's coming, what's missing (records, consent), and which patients are at risk of dropping.
- Catches scheduling mistakes before they happen - double-bookings, wrong location, wrong time.
- (Building) flags the same patient booked at two sites.

**Handling all the messages**
- Every reminder and confirmation goes out automatically, with guardrails so we never send the wrong message, to the wrong person, or at the wrong time.
- Incoming faxes and voicemails are read, sorted, and routed to the right person by AI.
- (Early/off by default) an AI voice agent that can phone patients to pre-screen them.

**Getting paid**
- Tracks what every study earns, what's been invoiced, what's been paid, and what's overdue.
- Shows profit per study after costs.
- A bot writes payments, invoices, and reconciliations back into the billing system for us (auto-marking an invoice fully-paid is built and tested but still switched off).

**Staying safe and compliant**
- Controls who can see what, and protects patient data.
- Keeps a full audit trail and auto-checks for data problems every morning.
- Every rule we care about has an automatic test behind it, so it can't quietly break.

**Not drowning in busywork**
- Dozens of scheduled jobs run throughout the day doing all of the above.
- A few "assistant" agents surface what needs attention (no-show risk, stale leads, money at risk) - but a human always approves before anything goes out to a patient or sponsor.

## The honest version (what's real vs. early)
- **Solid and running daily:** leads, scheduling, messaging, fax/voicemail triage, the finance/AR tracking, the compliance checks, the dashboards.
- **Built but held back on purpose:** the AI voice caller and the automatic collections/dunning emails - both work, both are switched off until we're 100% comfortable.
- **Still designing:** a couple of finance-validation pieces.

## What was hardest (the stuff worth comparing notes on)
- Making sure a write to another system (a booking, a payment) actually landed - not just that it said "OK." We learned this the hard way.
- Never letting a patient get a wrong or contradicting message.
- Doing reliable automation on top of a clinical system (CRIO) that has a limited API.

---

# Easy comparison - do you have this too?

Mark each **Yes / Building / No** for your site. Plain-English, no tech needed.

| What it does | Us | Your site |
|---|---|---|
| Auto-collect leads from website + ads | Yes | |
| Auto-check which studies a lead qualifies for | Yes | |
| Tell recruiters who to call first + track response speed | Yes | |
| Online booking + automatic visit reminders | Yes | |
| Win back no-shows / cancellations / unfinished forms | Yes | |
| One screen for the day's visits + what's missing | Yes | |
| Catch scheduling mistakes before they happen | Yes | |
| Catch the same patient enrolled at two sites | Building | |
| Automatic patient texts/emails with safety guardrails | Yes | |
| AI reads + routes incoming faxes and voicemails | Yes | |
| AI voice agent that pre-screens by phone | Built, off | |
| Track revenue / invoices / what's owed | Yes | |
| Profit per study | Yes | |
| Auto-update the billing system | Yes | |
| Auto-mark invoice paid | Built, off | |
| Automatic collections / dunning | Built, off | |
| Role-based access + patient-data protection | Yes | |
| Daily automatic data-quality checks + audit trail | Yes | |
| "Assistant" agents that flag what needs attention | Yes | |
| One shared design system across all our screens | Yes | |
| A patient website + a referring-doctor portal | Yes | |

**The point:** wherever our columns differ, that's either something worth sharing, or something one of us can stop building twice.