ai-intelligent-booking

Inspiration

• We’ve all wasted time emailing back-and-forth to schedule sessions. We wanted an AI that understands availability, preferences, and context, then just books the right slot—especially for counseling where matching and timing matter.

What it does

• Conversational booking assistant that guides users to the right counselor and time.
• Real-time availability lookup and conflict-aware scheduling.
• Magic-link authentication (no passwords).
• Automated email confirmations and reminders.
• Unified dashboards for users, counselors, and admins.
• Preference-aware matching (time, modality, counselor specialties).
• Admin tools for meeting link management and system health.

How we built it

• Frontend: Next.js App Router, React, Tailwind, shadcn/ui.
• Backend: Next.js API routes, Prisma ORM, PostgreSQL.
• AI: Tool-enabled agent via OpenRouter models, domain tools for availability, booking, and email.
• Auth: Magic-link flow and session middleware.
• Infra: Vercel deployment; Prisma migrations and seeding for fast bootstrap.
• Emails: React email templates for confirmations.

Challenges we ran into

• Normalizing availability and handling overlapping time zones.
• Balancing AI flexibility with deterministic booking rules.
• Mitigating LLM hallucinations via tight tool interfaces and validation.
• Rate limits and latency when orchestrating multi-step AI actions.
• Building a clean, role-based UX across user, counselor, and admin views.

Accomplishments that we're proud of

• End-to-end flow from conversation → booking → confirmation in minutes.
• Stable tool-use layer where the AI can only perform safe, auditable actions.
• Accessible, mobile-friendly UI with clear status feedback.
• Simple, secure login via magic links—no password fatigue.
• Modular design that’s easy to extend (new tools, new providers).

What we learned

• Clear tool contracts drastically improve AI reliability.
• Schema-first thinking (Prisma) speeds iteration and reduces bugs.
• Good UX turns complex AI flows into simple user decisions.
• Testing AI systems requires scenario-based validation, not just unit tests.

What's next for ai-intelligent-booking

• Calendar integrations (Google/Microsoft) with two-way sync.
• Payments, cancellation policies, and no-show handling.
• Smart reminders (SMS/email) and follow-up nudges.
• Better matching via embeddings and outcome feedback.
• Multi-tenant support for clinics; audit logs and analytics dashboards.
• Compliance hardening (PII minimization, encryption at rest/in transit).

Built With

• agentic
• ai
• nextjs
• openrouter
• postgresql
• prisma
• tailwindcss
• typescript