Landing Page
Enterprise-Grade Features Page
Sign-In / Authentication Page
Onboarding:Preferences Page
Main Recommendation Page
Smart Insights Panel (Trends & Peak Times)
Action Menu (Surprise Me / Date Night / Group / Budget)
Reservation / Booking Modal

Pick For Me :AI-Powered Decision Engine

Inspiration

We've all been there standing on a street corner, hungry and overwhelmed by endless restaurant options. You open Yelp, see 47 Italian restaurants, and suddenly you're paralyzed by choice. Decision fatigue is real, and it's exhausting.

The inspiration for Pick For Me came from a simple observation: traditional recommendation apps give you options, but they don't make decisions. They show you 50 restaurants and say "good luck choosing." What if AI could actually pick for you?

We realized that Yelp's new AI API represented a paradigm shift from recommendation engines to decision engines. This wasn't just about finding restaurants; it was about solving the fundamental problem of choice overload in our hyper-connected world.

What it does

Pick For Me is an AI-powered decision engine that eliminates choice paralysis by making confident, autonomous decisions for local experiences. Here's how it works:

Natural Conversation: Users describe what they want in plain English like "I want Italian food for dinner, under $50"
AI Decision Making: Instead of showing 20 options, our AI picks ONE with 92% confidence
Transparent Reasoning: We explain exactly why such as rating, price match, distance, cuisine compatibility
Autonomous Booking: One-click reservation through Yelp's booking system
Smart Insights:Location-aware insights surface what’s trending nearby, optimal visit times, and socially relevant signals like friends’ activity adding real-world context beyond static recommendations.
Action Panel: Users can choose intent-driven actions such as Surprise Me, Plan Date Night, Group Outing, or Budget Finder, instantly adapting the AI’s decision logic without additional input.
Universal Appeal: Works for tourists, locals, business travelers anyone tired of endless scrolling

The core innovation is decision delegation where users don't want more information, they want their time back.

How we built it

Architecture & Tech Stack

We built Pick For Me using a modern, scalable architecture:

Frontend:

Next.js 14 with App Router for optimal performance
TypeScript for type safety and developer experience
Tailwind CSS with custom neo-brutalism design system
React Hook Form + Zod for robust form validation
Firebase Authentication for user management

AI Integration:

Yelp AI API v2 for conversational intelligence
Yelp Places API for business data enrichment
Yelp Reservations API for automated booking
Custom decision engine with weighted scoring algorithm

Key Technical Innovations:

1. Decision Engine Algorithm

We developed a sophisticated scoring system that weighs multiple factors:

$$\text{Decision Score} = 0.3 \times \text{Rating} + 0.25 \times \text{Price Match} + 0.2 \times \text{Distance} + 0.15 \times \text{Cuisine} + 0.1 \times \text{Popularity}$$

2. Confidence Calculation

Since Yelp's API doesn't provide confidence scores, we engineered our own:

$$\text{Confidence} = \frac{\text{Top Score} - \text{Second Score}}{\text{Top Score}} \times 100\%$$

3. Context Preservation System

We built stateful conversation management that tracks:

User preferences across sessions
Location context and travel patterns
Previous decisions and feedback loops
Real-time availability and booking status

Challenges we ran into

1. The Decision vs Recommendation Gap

Challenge: Yelp's AI API excels at understanding queries and providing recommendations, but lacks built-in decision-making logic for autonomous selection.

Solution: We built a custom decision engine that bridges this gap, turning Yelp's recommendations into confident commitments.

2. Trust and Confidence Communication

Challenge: Users need to trust AI decisions, but how do you communicate confidence in autonomous choices?

Solution: We developed transparent confidence metrics (85-98% range) and explainable reasoning that shows users exactly why the AI made its choice.

3. Context Preservation Complexity

Challenge: Multi-turn conversations require sophisticated state management that isn't automatically handled by the API.

Solution: We implemented a conversation context system that maintains user preferences, location data, and decision history across interactions.

4. Booking Integration Friction

Challenge: Moving from AI recommendation to actual reservation traditionally requires multiple steps and user actions.

Solution: We created seamless one-click booking that integrates Yelp's Reservations API with our decision engine, reducing the path from "I'm hungry" to "table booked" to under 60 seconds.

5. Neo-Brutalism Design Implementation

Challenge: Creating a bold, accessible design system that stands out while maintaining usability.

Solution: We developed a comprehensive neo-brutalism component library with 4px black borders, bold shadows, and interactive feedback that makes the AI's confidence visually tangible.

Accomplishments that we're proud of

1. True Agentic AI

We didn't just build another recommendation app but we created an AI that actually makes decisions and takes actions. This represents the future of AI assistants.

2. Solving Real Problems

We addressed the "paradox of choice" that affects millions of Yelp users daily. More options don't always mean better decisions.

3. Technical Innovation

Custom decision engine with mathematical confidence scoring
Seamless multi-API integration (AI, Places, Reservations)
Sophisticated conversation state management
Novel neo-brutalism design system

4. Universal Appeal

Our solution works for everyone:

Tourists who don't know the area
Locals tired of the same spots
Business travelers who need quick decisions
Anyone experiencing decision fatigue

5. Commercial Viability

Pick For Me has clear business value for Yelp:

Increases conversion from browse to book
Reduces user drop-off from decision fatigue
Demonstrates cutting-edge AI capabilities
Creates stickier user engagement

What we learned

1. AI Needs Decision Frameworks

We discovered that conversational AI is powerful for understanding intent, but needs structured decision frameworks to take autonomous actions. The gap between "understanding" and "deciding" is where the real innovation happens.

2. Confidence is Key to Adoption

Users will delegate decisions to AI, but only if they trust it. Transparent reasoning and confidence metrics are essential for autonomous AI systems.

3. Context is Everything

Location-aware AI isn't just about GPS coordinates, it's about understanding user context, travel patterns, and situational needs. A tourist and a local have completely different decision criteria.

4. Design Affects Trust

Our neo-brutalism design system isn't just aesthetic - the bold, confident visual language reinforces the AI's decision-making confidence. Design and functionality are inseparable in AI interfaces.

5. API Orchestration is Complex

Integrating multiple APIs (AI, Places, Reservations) requires careful orchestration and error handling. The user experience depends on seamless coordination between services.

What's next for Pick For Me

Short-term Enhancements (Next 3 months)

Multi-day Trip Planning: Extend from single decisions to complete itineraries
Voice Interface: Natural speech input for hands-free decision making
Social Integration: Share decisions and get group consensus
Advanced Personalization: Machine learning from user feedback

Medium-term Expansion (6-12 months)

Global Scaling: Expand beyond US to international markets
Category Expansion: Hotels, attractions, transportation, events
Business Intelligence: Analytics dashboard for restaurants and venues
API Platform: Allow other developers to integrate our decision engine

Long-term Vision (1-2 years)

Predictive Decisions: AI that anticipates needs before users ask
Ecosystem Integration: Connect with calendar, weather, traffic, social plans
Enterprise Solutions: Corporate travel and event planning
AI Agent Network: Multiple specialized AI agents for different decision domains

Technical Roadmap

Real-time Adaptation: Dynamic re-planning based on changing conditions
Advanced ML Models: Custom recommendation models trained on decision outcomes
Blockchain Integration: Decentralized reputation and review systems
AR/VR Interfaces: Immersive decision-making experiences

Pick For Me represents more than just a travel app it's a new paradigm for human-AI interaction where AI doesn't just inform, but decides. We're building the future where technology doesn't give us more choices, but makes better choices for us.

Stop Choosing. Start Living. 🌟