Github: https://github.com/Vikhyat05/Sarvam_AI_Assignment (Shared with : atishay@sarvam.ai)

Q1. Comprehensive Use-Case Document

Long-Term Goal

/

Position FoodieSpot’s AI assistant as the go-to entry point for dining decisions—boosting table-fill rates, increasing average party size, and capturing richer customer data across all locations—while cutting labor costs and freeing staff to deliver an exceptional in-house experience. Additionally, the assistant can evolve to allow customers to place orders directly from their tables, eliminating wait times and further enhancing convenience and satisfaction.

Success Criteria

Use Case

City diners message FoodieSpot on WhatsApp, web chat, or voice (“Hey FoodieSpot, table for four tonight near Indiranagar”). The AI reads their intent, checks live capacity, recommends the best-fit venue/section (e.g., “rooftop patio with live music”), secures a table, and pushes a confirmation plus add-ons (birthday cake, valet). Minutes before the meal it can adjust timing, upsell specials, and handle cancellations. The same flow works across all FoodieSpot branches, turning every inquiry into a data-rich reservation and follow-up revenue opportunity.

Key Steps – End-User Flow

1 Greet & Capture — Bot greets, parses date, party size, area, cuisine. 2 Search & Recommend — Live query to seating DB ➜ shortlist best locations/sections. 3 Confirm Details — Ask missing info (name, phone, special requests). 4 Book — Create reservation, send confirmation plus calendar link. 5 Pre-Visit Engagement — Reminder + upsell (menu preview, add dessert). 6 Modify / Cancel — User can reschedule via same channel. 7 Post-Meal Feedback — 1-tap rating; incentives for review.

State Transition Diagram

[] --> Idle Idle --> Capturing: userMessage Capturing --> Recommending: gotPreferences Capturing --> Capturing: partialPreferences Recommending --> Confirming: suggestionAccepted Recommending --> Capturing: changePrefs Confirming --> Booking: detailsComplete Confirming --> Capturing: missingInfo Booking --> Booked: success Booking --> Error: failure Booked --> Modifying: userChange Modifying --> Booked: updateSuccess Booked --> FollowUp: afterDining FollowUp --> []