AI Receptionist: How We Built an AI–Human Recruitment System for a Kidney Disease Clinical Study
Scaling patient outreach without sacrificing consent, trust, or compliance.
A man named Ian Ho approached me with a very specific, and very sensitive, challenge.
They were running a kidney disease clinical study in partnership with doctors.
The doctors had already opted their patients in and granted permission to contact them.
The problem wasn’t whether they could call patients.
The problem was how to do it properly, at scale, without burning trust.
Clinical recruitment lives at the intersection of:
- Compliance
- Human empathy
- Operational efficiency
Most systems fail because they lean too hard on manual calling or go fully robotic and impersonal.
We designed something different.
The Core Objective
The goal was simple to state, but hard to execute well:
Contact consented patients, pre-screen them, and schedule doctor appointments — while preserving a human experience and minimizing staff workload.
This meant:
- No spammy robocalls
- Clear disclosure
- Respect for patient preferences
- Smooth handoffs between AI and humans
The Big Idea: Human-AI Collaboration, Not Replacement
Instead of replacing staff, we built a system where:
- AI handles volume, timing, and repetition
- Humans handle nuance, edge cases, and trust-building
Every step is logged.
Every outreach is intentional.
Every patient can choose how to engage.
High-Level Architecture
The system combines:
- Twilio – calling, SMS, email infrastructure
- Bland.ai – human-like AI caller (inbound + outbound)
- Slybroadcast – compliant voicemail drops
- TidyCal – appointment scheduling
- Smartsheet + Google Sheets – CRM and operations layer
- OpenAI – conversational intelligence and screening logic
Think of it as a clinical recruitment control tower.
End-to-End Workflow
1. Lead Acquisition & First Contact (Text, Email, Voicemail)
Trigger
A lead signs up via a Facebook campaign form for the kidney disease study:
- Phone number
- Preferred appointment date & time (if provided)
Automated Actions
Personalized SMS + email sent immediately:
- Welcome message
- Brief explanation of the study
- TidyCal link if no appointment was selected
Ringless voicemail drop (using a realistic human voice):
“Hi, this is [Your Name]. Thanks for signing up! When would be the best time for someone from our team to give you a quick call?”
Follow-Up Logic
- Day 3: Voicemail drop repeated
- Day 7: Final voicemail emphasizing urgency and benefits
No pressure.
No harassment.
Just structured persistence.
2. Conditional Engagement Flow (Human + AI)
Trigger
The lead responds or consents to contact.
Primary Path: Human First
- A team member manually calls at the preferred time.
If the call is missed
- After a defined delay (e.g. 1 day), the Outbound AI Caller Bot takes over.
AI Caller Capabilities
- Clearly identifies itself
- Asks for consent
- Begins pre-screening if allowed
If AI call is missed
Automatic SMS + email sent:
- “We tried reaching you…”
Options to:
- Reply by text
- Book via TidyCal
- Call back directly
3. AI-Powered Pre-Screening & Doctor Scheduling
(Text + Email focused)
Trigger
Lead engages with the AI Caller Bot.
AI Actions
Asks structured pre-screening questions:
- Availability
- Eligibility indicators
- Willingness to proceed
If the lead qualifies
- AI schedules an appointment directly with the clinical trial doctor
Sends confirmation SMS + email with:
- Date
- Time
- Instructions
If eligibility is unclear or complex
- The system flags the case
- Human staff are notified for personalized follow-up
This prevents AI from making judgment calls it shouldn’t.
4. Callbacks & Inbound Handling (Human or AI)
Trigger
A lead calls back after receiving a voicemail, SMS, or email.
If a human receptionist is available
They answer with a compliant script:
“Hi [Name], thanks for calling back. I can guide you through a quick pre-screening and help schedule your appointment if you qualify.”
If no human is available
The Inbound AI Caller Bot answers transparently:
“I’m an AI assistant designed to make this process smooth and convenient. If you’d prefer a human, just let me know.”
The lead always has a choice.
5. Post-Screening Follow-Up (SMS + Email)
Trigger
Lead completes pre-screening but does not schedule.
Actions
- Polite reminder via SMS + email
- FAQs addressing common concerns
Option to:
- Schedule later
- Request a callback
- Speak to a human
The AI Caller Bot can re-engage if appropriate.
Automation Enhancements Behind the Scenes
Calendar Sync
- Vacant doctor slots are crawled and reserved dynamically
Appointment Creation
- Events auto-inserted and synced across calendars
Inbound & Outbound AI Callers
- Human-like voice, compliant disclosure
Voicemail Drops
- Used only where consent and DNC checks allow
CRM & Lead Tracking
- Every call, SMS, email, voicemail logged automatically
Nothing slips through the cracks.
Why This Works
- Patients aren’t overwhelmed
- Staff aren’t burned out
- Compliance is respected
- Conversion rates improve
- Doctors’ time is protected
This is automation with restraint — designed for healthcare realities, not startup demos.
Other Optional Automations |
|
|
|
|
Final Thoughts
Clinical recruitment doesn’t need more people dialing phones all day.
It needs systems that respect humans on both sides of the line.
That’s what we build at WunderWaffen:
AI systems that scale operations without stripping away trust.
