Launching today
Pickify - AI Gift Finder
Stop overthinking gifts. Let AI find the perfect match
6 followers
Stop overthinking gifts. Let AI find the perfect match
6 followers
Stop spending hours browsing Amazon, hoping to find the perfect gift. Pickify uses AI to do the hard work for you. Answer a few questions about the recipient (age, interests, occasion, budget) and get 4 personalized gift recommendations instantly. Each comes with direct purchase links. What makes it special: • Learning system that tracks what gifts work best • Works globally (Amazon, KSP, more coming) • Free, no signup required • 60 seconds from start to recommendations
@ran_kopchovski
What kind of data do you collect? I'm always worried about privacy with AI tools.
@thomas_bentov
Totally valid concern! We're very minimal with data:
✅ What we collect:
• Quiz answers (recipient type, interests, etc.)
• Anonymous session IDs
• Which gifts get clicked (for learning system)
❌ What we DON'T collect:
• No names, emails, or personal info
• No tracking across websites
• No selling data to third parties
Everything is anonymous and only used to improve recommendations. You can use it without even creating an account.
Full transparency: we do earn affiliate commission if you buy through our links, but that doesn't influence what we recommend (the AI optimizes for click-through rate, not commission rate).
👋 Hey Product Hunt! I'm Ran Kopchovski, solo maker.
I built Pickify because I spent way too long trying to find the perfect birthday gift for my mom last year - 5 hours of browsing Amazon, still ended up with something generic. I thought there has to be a better way.
So I built an AI that asks the right questions and gives you confident, personalized gift recommendations in 60 seconds.
🎯 What makes Pickify different:
• Learning system - it tracks which gifts actually get clicked for different people/occasions and optimizes over time (we're in learning phase now, collecting data!)
• Multi-affiliate support - shows the best store for your country (Amazon / Aliexpress / KSP ,more stores coming)
• Just 4 focused recommendations - no decision paralysis from 50 options
• Free forever, no signup required
🔧 Tech stack:
• Claude Sonnet 4 (Anthropic) for AI recommendations
• React + Vercel + PostgreSQL with Prisma
🚀 What's next (based on early feedback):
• Gift reminders for upcoming birthdays
• Shareable results ("See what AI picked for my mom!")
• Group gift mode (multiple people suggest for one person)
I'd love your feedback! What features would make this more useful for you?
Try it now: pickifyai.com 🎁
Ask me anything about the AI, the learning system, or gift-giving psychology!
Can you share how you find the best gift? Can you share the general algorithm?
@noam_cohen91 Great question!
Happy to pull back the curtain 🎯
The algorithm works in 3 phases:
Phase 1: Data Collection (The Quiz)
We gather 7 key factors:
• Recipient type (mom, partner, friend, etc.)
• Age range (affects product appropriateness)
• Occasion (birthday vs holiday = different expectations)
• Budget (hard constraint - never exceed this)
• Interests (tech, fashion, sports, etc.)
• Style preferences (practical, fun, luxury, personal, experience)
• Additional context (free text for nuance)
Phase 2: AI Recommendation (Claude Sonnet 4)
We send these factors to Claude AI with a carefully engineered prompt that:
• Emphasizes the specific combination (e.g., "tech-loving mom, 50-64, birthday, $50-100")
• Provides market-specific pricing guidelines
• Enforces strict budget constraints (this was the hardest part to get right!)
• Requests structured JSON output (4 gifts with title, description, reason, price, emoji)
The AI considers:
✅ What products match the interests
✅ What's age-appropriate
✅ What fits the occasion context
✅ What's within budget
✅ Cultural preferences (Israeli vs US gift-giving is different!)
Phase 3: Learning & Optimization (Our Secret Sauce)
This is where it gets interesting! We track:
• Which gifts get clicked (Buy Now button)
• For which quiz combinations (recipient + occasion + interests)
• Which affiliate store performs better per country
After collecting 1000+ clicks for a specific combination (e.g., "tech-loving mothers, birthdays, Israel"), the system enters "optimization mode" and prioritizes recommendations that historically perform better.
Example:
If we learn that "tech moms aged 50-64 in Israel" click on tablets 3x more than smartwatches, the AI will prioritize tablet recommendations for that profile.
Right now we're in "learning phase" (50/50 split between affiliates, randomized order) collecting this data. The system gets smarter over time!
Does that answer your question? Happy to dive deeper into any part! 🤓
Have you tried it yet? Curious what you think of the results!