Challenge 2

bySoban

Challenge 2: AI Service Orchestrator for Informal Economy Challenge Overview Informal economy, including plumbers, electricians, tutors, beauticians, and home service providers, operates largely through: ● WhatsApp messages ● phone calls ● informal referrals This results in: ● inefficient service matching ● missed opportunities ● lack of automation ● poor user experience At the same time, users struggle to find: ● reliable services quickly ● availability in real time ● trusted providers nearby Problem Statement Build an Agentic AI System that automates the end-to-end lifecycle of a service request — from user intent to booking and follow-up. Your system must: 1. Understand user service requests (in natural language) 2. Identify relevant providers using location/context 3. Select or recommend the best provider 4. Simulate booking and confirmation 5. Handle follow-up interactions 6. Show complete reasoning and workflow execution Mandatory Requirement: Google Antigravity Teams MUST use Google Antigravity as the core platform to: ● orchestrate agent workflows ● manage multi-step reasoning ● integrate tools (Maps, Search, APIs) ● execute actions (booking, notifications, etc.) Use of external LLMs is allowed, but Antigravity must be central to system logic and orchestration. Example User Scenario User input (Roman Urdu / Urdu / English): “Mujhe kal subah G-13 mein AC technician chahiye” Expected Output Service Request: AC Technician Location: G-13 Time: Tomorrow morning Recommended Provider: Ali AC Services (2.1 km away) Reasoning: Closest available provider with high rating Simulated Booking: - Slot booked: 10:00 AM - Confirmation sent Follow-up: Reminder scheduled 1 hour before appointment System Requirements 1. Intent Understanding ● Process natural language input ● Support: § Urdu § Roman Urdu § English ● Extract: § service type § location § time 2. Provider Discovery ● Use: § mock dataset OR § Google Maps / Places APIs ● Identify: § nearby providers § service category match 3. Matching & Ranking ● Rank providers based on: § distance § availability § rating (simulated or real) ● Provide clear reasoning for selection 4. Decision & Recommendation ● Select best provider OR show top options ● Explain decision in simple terms 5. Action Simulation (CRITICAL REQUIREMENT) System must simulate: ● booking confirmation ● provider assignment ● scheduling Simulation can include: ● updating a mock booking system ● creating a confirmation message ● writing to a database/spreadsheet ● generating a booking receipt 6. Follow-Up Automation ● Simulate: § reminders § status updates § completion confirmation 7. Agentic Workflow (MANDATORY) System must demonstrate: ● multiple agents OR structured reasoning pipeline ● planning → decision → action → follow-up ● traceable logs of: § decisions § tool usage § action execution Deliverables 1. Working Prototype with Mobile App (MUST) and Web App (Optional) 2. Demo Video (3–5 minutes) Must clearly show: ● user input ● system understanding ● provider matching ● booking simulation ● follow-up workflow 3. Agent Trace / Logs ● reasoning steps ● agent interactions ● action execution logs 4. Documentation (README) Include: ● system architecture ● how Antigravity is used ● APIs/tools used ● assumptions and limitations Evaluation Criteria 1. Use of Google Antigravity — 25% ● Core orchestration handled via Antigravity ● Effective use of tools (Maps, APIs) ● Demonstrates planning + execution 2. Agentic Reasoning & Workflow — 20% ● Multi-step reasoning ● Logical flow from request → decision → action ● Evidence of autonomy 3. Matching Quality & Decision Logic — 20% ● Relevant provider selection ● Clear ranking criteria ● Strong reasoning behind decisions 4. Action Simulation & Execution — 15% ● Booking process realistically simulated ● Clear system state change (confirmation, scheduling) ● End-to-end workflow demonstrated 5. Technical Implementation — 10% ● Clean architecture ● API/tool integration ● Robust handling of edge cases 6. Innovation & UX — 10% ● Creative approach ● Intuitive interface ● Clear and engaging demo Important Guidelines ● This is NOT a simple listing or booking app ● Focus on agentic automation, not UI complexity ● At least one booking must be simulated end-to-end ● Must demonstrate reasoning + decision-making ● Use mock data if real APIs are unavailable ● Avoid use of real personal/sensitive data