Safari AI

01The Problem

Safari operators, DMCs and agents build itineraries by hand, then track leads, quotes, bookings and payments across spreadsheets, email and chat.

Slow quotes lose bookings, and scattered records make it easy to miss a payment or an unconfirmed booking. Travellers planning alone face the same blank page with less local knowledge.

02The Approach

I split the product by audience at signup. Travellers get the AI planner. Professionals get the planner plus a CRM built around trips.

I used Google Gemini through Genkit for itinerary generation, and Firebase for auth and data, so I could ship as a solo builder without running servers.

03The Solution

Safari AI is live at safariai.app.

• Free-text trip requests, such as a 10-day honeymoon safari, become day-by-day itineraries that users can refine. Gemini also generates trip images.
• A travel CRM with a lead and trip pipeline, quotes, payments, a bookings calendar, tasks tied to trips, suppliers, client records and a client portal.
• A Copilot assistant inside the CRM, and a dashboard that flags overdue tasks, overdue payments, unconfirmed bookings and departures in the next 14 days.
• Firebase Auth with email or Google sign-in, Firestore for data, and Stripe Checkout and the customer portal for subscriptions.
• Plan tiers for travellers and teams, with org branding, team members, custom fields and white-label on higher tiers.

04The Why

Firebase and Vercel keep operations close to zero, so my time goes into the product.

Separate traveller and professional modes let one AI engine serve two markets without cluttering the traveller experience.

A seeded Kenya sample pipeline means sales demos always show realistic data, never empty screens.

05The Result

Safari AI is live and taking signups, with a free CRM trial for professionals. It is on show to operators at the Magical Kenya Travel Expo in Nairobi in October 2026.

06Stack

Visit site