Appointment Rescue

**Appointment Rescue** doesn't just send reminders—it uses CALL-E voice AI to autonomously call customers, negotiate new times, and turn unconfirmed appointments into recovered revenue.

Comment

Inspiration

Every year, healthcare clinics and appointment-based businesses lose thousands of dollars to unconfirmed or missed appointments. Traditional SMS and email reminders notify patients, but they rarely resolve scheduling conflicts or recover lost revenue. We built Appointment Rescue around a simple core idea: Don't just remind people about appointments. Rescue appointments that would otherwise be lost.

What it does

Appointment Rescue is an autonomous AI agent that identifies at-risk appointments, places outbound phone calls using CALL-E, conducts natural voice conversations, and recovers appointments by:

  • Confirming attendance for upcoming slots,
  • Rescheduling to available alternative time slots dynamically,
  • Recording cancellation reasons when customers relocate or decline,
  • Safely escalating complex requests (e.g., medical questions) to clinic staff,
  • Updating workflow metrics and revenue retention statistics in real time.

How we built it

  • Voice AI Engine: Integrated CALL-E's TypeScript SDK (@call-e/calle) and CLI with structured JSON result extraction schemas.
  • Backend Architecture: Node.js, Express, TypeScript, and SQLite for deterministic slot matching, audit event logging, and telemetry attribution (CALLE_SOURCE=skills_sh).
  • Frontend Dashboard: React 18, Vite, Tailwind CSS, and Lucide icons featuring KPI summary cards, an AI Rescue Queue, a Live Activity Stream, and Human Escalation Queue.
  • Dual Execution Modes: Includes a Live Calls mode for real phone calls and a Demo Mode simulating 5 key hackathon scenarios.

Challenges we ran into

  • Conversational Guardrails: Ensuring the AI voice agent negotiates time slots strictly within provided clinic availability without inventing times or giving medical advice.
  • Structured Schema Extraction: Reliably parsing spoken phone conversations into structured JSON outcomes (confirmed, rescheduled, human_review, cancelled) to drive automated database updates.
  • Real-Time UI Telemetry: Translating live call states (Connecting → Conversation → Outcome) into an interactive transcript and progress indicator.

Accomplishments that we're proud of

  • End-to-End Autonomous Workflow: Clicking "Rescue Appointments" initiates real CALL-E phone calls and automatically updates appointment statuses without human intervention.
  • Safe Agent Autonomy: Successfully implementing human escalation when customers ask out-of-scope questions, demonstrating responsible AI behavior.
  • Polished SaaS Experience: Delivering a production-grade UI with financial impact analytics demonstrating revenue retained versus baseline loss.

What we learned

  • Voice AI phone calls dramatically outperform static text reminders for conversational scheduling and capturing customer intent.
  • Setting strict schema boundaries and explicit behavioral guardrails for voice agents is essential for safe real-world deployment.

What's next for Appointment Rescue

  • EHR & Calendar Integrations: Direct two-way sync with Google Calendar, Outlook, Cal.com, and clinic management software (Epic/Cerner).
  • Multi-Lingual Voice Outreach: Expanding CALL-E language models to support multi-lingual patient bases.
  • Inbound AI Receptionist: Handling inbound customer callbacks to reschedule or answer clinic inquiries 24/7.

Built With

Share this project:

Updates

Submission history