AI leads scheduler

Inspiration

The inspiration came from our own experience running service-based projects and realizing that the scheduling process could be smarter, faster, and more human-like.

What it does

We wanted to create an AI-powered scheduling assistant that could:

• Communicate naturally with clients via chat or email.
• Understand appointment requests, preferences, and constraints.
• Automatically reserve slots in a business calendar.
• Send reminders and updates without human intervention. ## How we built it
• Framework Integration
  We used our self made PHP framework MVCA as the base, ensuring the scheduling logic fits seamlessly into existing websites.
• AI Communication Layer
  Integrated with a conversational AI API to process natural language from clients.
  The AI extracts key details such as:
  • Desired date/time
  • Service type
  • Special requests

1. Calendar & Booking Engine
   Connected to Google Calendar API and an internal MVCA scheduling module to manage availability.

2. Automation Logic
   If a request meets availability conditions: $$ \text{Booking} = \text{ClientRequest} \cap \text{AvailableSlots} $$ …then the system reserves it instantly and confirms with the client.

3. Reminder System
   Sends notifications via email/SMS automatically before the appointment.

Challenges we ran into

• Context Retention: Teaching the AI to remember client preferences within a conversation.
• Error Handling: Managing cases where the AI misunderstands requests and ensuring no double bookings occur.

Accomplishments that we're proud of

• Built a fully functional AI scheduling assistant, ready to integrate into real MVCA-powered websites.
• Achieved seamless conversation flow where clients can book appointments without any manual intervention.
• Created a modular architecture that can easily adapt to other industries beyond scheduling (support, sales, onboarding).
• Solved tricky double-booking and time zone issues, ensuring accuracy and reliability.
• Delivered a working live demo with real-time AI responses and calendar updates.

What we learned

• How to combine natural language processing with strict business logic without losing conversation quality.
• Best practices for calendar API integrations and maintaining data consistency.
• Techniques for error-proof AI workflows, including fallback responses and client confirmation steps.
• The importance of clear UX design in AI-driven systems so users understand what’s happening at each step.
• That even a small MVP, when focused on solving a real-world pain point, can deliver massive business value.

What's next for AI leads scheduler

• Multi-Channel Support: Extend beyond web chat to SMS, WhatsApp, and voice calls.
• Advanced Client Profiling: Let AI learn client preferences over time to make proactive suggestions.
• Team Scheduling: Support multiple staff calendars and auto-assign based on availability and skill set.
• Analytics Dashboard: Provide insights on booking trends, peak times, and client retention.
• Industry Templates: Create pre-configured scheduling models for specific industries like healthcare, fitness, and consulting.
• Self-Hosting Option: Allow businesses to run the AI scheduler on their own servers for data privacy compliance.

Instructions

1. Download or clone staging branch: https://github.com/NovDmytro/mvca/tree/MVCA-HKTN-1
2. Run "docker compose up"
3. Insert your OpenAi API key at mvca/app/config/config.php Line 42: 'openAIKey'=>'',
4. Run demo socket here: http://localhost:2121/Samples.AI.worker/ (keep in mind that in production this socket will run constantly, but now, it's a test example so socket will be alive only for the 600 seconds)
5. Chat with AI to Scheldule your Tv mounting service here: http://localhost:2121/Samples.AI.main/

Built With

• javascript
• mariadb
• mvca
• php