Case Study · Conversational AI · Healthcare
FamilyCheck
An AI voice agent for the people you love, in private beta
FamilyCheck places warm daily medication check-in calls to elderly family members and keeps caregivers informed after every call.
Under the hood, FamilyCheck is a fully autonomous voice agent Modeo designed, built, and operates in production: consent-gated outbound calls, retrieval over each patient's medications, real-time tool use, and multi-channel notifications on a Vapi, Twilio, and Resend stack.
What is FamilyCheck?
FamilyCheck is a consumer product, live at family-checker.com. A caregiver enters a loved one's name, medications, and preferred call times. FamilyCheck then calls daily with a warm voice, asks about the medication, and sends the outcome to the caregiver over WhatsApp or email in seconds. It works on any phone: no app, no smartphone, no internet.
For Modeo, it is also the proof behind everything we tell enterprise buyers: a regulated-context voice agent that runs autonomously in production, built and operated by the same team you would engage.
What did Modeo build?
The agent runs on Vapi and Twilio for telephony and voice orchestration, with OpenAI models for conversation and ElevenLabs for the voice, and Resend for caregiver notifications. Every call is an autonomous, multi-turn conversation with real-time tool use: the agent retrieves the patient's actual medication list mid-call and logs a structured outcome when the call ends.
- Consent-gated outbound calling: the agent calls the patient first, and daily reminders begin only after verbal agreement
- Retrieval over each patient's medications, so the conversation names the right drug at the right time
- Structured outcome logging per call, delivered to caregivers over WhatsApp or email
- Fail-closed webhook verification across Vapi, Twilio, and Stripe
How does a voice agent stay safe in a regulated context?
Healthcare tolerates no improvisation, so the system is built consent-first and eval-driven. Nothing about the agent's behavior changes casually: prompt changes are scored against recorded scenario suites before they ship, and call flows that matter, consent, sign-off, escalation, are covered by dedicated eval suites.
The human stays in the loop by design. Caregivers see the outcome of every call, and the patient can opt out verbally at any time.
Why it matters for enterprise buyers
FamilyCheck is a voice agent in private beta with its first consented patient, in a healthcare context, operated by the person who built it. It is the evidence for how Modeo specifies, tests and runs an agent: a 53-scenario adversarial test suite and a judge model that scores each call.
Where this pattern applies, and what a Review tests
Where the pattern applies
- Appointment confirmation and rescheduling calls
- Insurance renewal and missing-document calls
- Delivery and service-window confirmation
- Post-visit and post-service check-in calls
What a Reliability Review tests on a system like this
- Does every reported concern reach the caregiver with the symptom named?
- Are medical-advice and off-topic requests declined and logged with the right status?
- Does the agent stay calm and avoid emergency language when the patient is not in danger?
- Is anything private or internal ever spoken aloud?
Get in Touch
Start with the agent stuck in pilot.
Bring one agent and one workflow. In 4 weeks you have a pass rate on your own traces, a test suite in your CI, and a written go or no-go.
Recommended first step
A 30-minute call on where your agent stands and whether a Reliability Review fits. Prefer email? Write to [email protected] instead.
Book a review callNo meeting yet? Request a written proposalBuilt with leading technology