DEV Community

Cover image for The First AI-Powered URL Shortener - Shrinkify
Jordan Tingling
Jordan Tingling

Posted on

The First AI-Powered URL Shortener - Shrinkify

What does he mean by AI-powered URL Shortener?

Hey Guys, I'm Launching Shrinkify today and figured this community i've been apart of for a longtime would appreciate the honest behind-the-scenes.

I had an issue with how things were

I was paying $29/mo to Bitly. For what? A database entry and a 301 redirect. Maybe some basic charts if I was lucky.

The idea wondered in my head for a while, that shortening urls don't need to be that expensive and most of the current solutions only shorten links, people still left with many questions unanswered. I wanted something that could fill in the blanks to the questions I had and many others do about how their links are actually performing? what they could be doing differently? and I wanted an solution that was more generous but I couldn't find one so... I built my own.

What makes it different (the technical bits)

Instead of just another CRUD app with redirects, I added AI that's actually useful:

Conversational analytics – Ask questions in plain English:

"Which link got the most clicks this week?"
"Is my landing page still up?"
"When's the best time to post my next link?"
Enter fullscreen mode Exit fullscreen mode

shrinkify-ai

No GraphQL queries. No SQL. Just ask and get answers with real data. Your data! your links/urls! your campaigns/products/videos! all these data sources will feed metrics into Shrinkify where users can derive actionable insights from to actually make decisive decisions.

Under the hood:

  • React 18 + TypeScript + Vite
  • Supabase for backend (Postgres + real-time subscriptions)
  • OpenAI API for conversational interface along with all the fancy models (ChatGPT, Claude, Gemini, and more)
  • ML models for predictive analytics
  • Edge functions for AI health monitoring

shrinkify-ai

Other features devs will appreciate:

  • 🔮 Predictive analytics (7/30/90-day forecasts)
  • 🏥 Automated health monitoring with alerts
  • ⚡️ AI slug generator (because naming is the hardest problem)
  • 🔌 High-speed API access on premium is coming soon also
  • 📊 CSV exports for your own analysis

Pricing (actually fair)

  • Free: 25 links, 10K clicks, basic AI
  • Pro: $9/mo (1k links, 100K clicks)
  • Premium: $99/mo (unlimited links, 1M clicks/mo, and all AI features)

Compare that to Bitly's $29/mo for basic features.

The tech stack

Frontend: React 18, TypeScript, Tailwind CSS, shadcn/ui, kibo UI
Backend: Supabase (Postgres + Edge Functions)
AI: OpenAI GPT-4 API, Gemini, Claude
Analytics: Custom ML models
Deployment: Vercel
Payments: Stripe
Enter fullscreen mode Exit fullscreen mode

Open to questions about any of the architecture decisions.

Why I'm posting here

I'm indie building this. No VC money. No team. Just me and i'm having fun.

The code isn't perfect. There are probably bugs. But it's honest work and it actually solves a problem I had.

Would love feedback from this community:

  • Is the AI useful or gimmicky?
  • What features would you actually use?
  • What would make you switch from your current solution?

Try it out, break it, tell me what sucks: shrinkify.co

I'll be around to answer questions 🙏

⭐️ Shrinkify's Product Hunt Link: AI URL shortener that doesn't feel like a scam

shrinkify-ai

P.S. - If you're curious about how I implemented the conversational AI or the predictive models, happy to write a deep-dive post.

Top comments (12)

Collapse
 
shemith_mohanan_6361bb8a2 profile image
shemith mohanan

This is a refreshing take — most “AI-powered” products slap a chatbot on top, but conversational analytics for link performance is actually a meaningful upgrade. The ability to ask “When’s the best time to post?” or check link health like a status page feels like a real unlock vs dashboards people never read.
Curious: in your early tests, which AI feature are users finding the most value in — predictive forecasts or the natural language insights?

Collapse
 
jordan_t profile image
Jordan Tingling

I’d say based on my responses so far it’s been the natural language insights more! @shemith_mohanan_6361bb8a2 because the predict forecasts, analytics and health monitor for each link are data sources for the AI link assistant so the more data being ingested from these sources is the better insight the assistant will provide!

Collapse
 
akshatraj00 profile image
Akshat Raj

Wishing you joy, success, and all the happiness life can bring. May your dreams come true and each day be filled with positivity. Best wishes always!

Collapse
 
jordan_t profile image
Jordan Tingling

Thank you @akshatraj00 !! 🙌 same to you!

Collapse
 
varshithvhegde profile image
Varshith V Hegde

App looks well rounded ...
All the best with the launch

Collapse
 
jordan_t profile image
Jordan Tingling • Edited

Thank you! @varshithvhegde 🙌

Collapse
 
codemonster240 profile image
Andrew McSillyone

Slight problem: AI APIs are expensive, and bots can easily abuse them. Do you have something that prevents abuse and doesn't run up your API bill? I have come across this problem a ton with my AI stuff.

Collapse
 
jordan_t profile image
Jordan Tingling

Hey @codemonster240 thats a great question! The solution is having hard caps and rate limits in place with cooldown mechanisms. Each of the AI services that Shrinkify provides has different limits/daily caps and cooldowns enforced depending on the subscription type. So that even with abuse or unexpected usage, you’ll have better control of costs and users still receive the services and functionality they were promised

Collapse
 
alifar profile image
Ali Farhat

Vite 🙌 🙌 🙌

Collapse
 
jordan_t profile image
Jordan Tingling

⚡️
25 users already. that's 25 people who trusted a guy who launched at midnight. thank you, you legends!

happy-25

Collapse
 
pranav_thampi_801e8cabe86 profile image
Pranav Thampi

I'd love to see the deep dive on how you implemented the AI for conversation as well as predictive analysis while keeping costs low and preventing abuse

Collapse
 
jordan_t profile image
Jordan Tingling

Fairly straightforward actually = The AI assistant simple derives context and data points about current links and how they have performed since being made - to present moment. Then we conduct an analysis from these data sources to make a predictive forecast and best recommendations for growth that are specific to your links, your traffic patterns, # of clicks on your links, best time to posts/share your links based on your past engagement patterns and other factors. To best assist is providing you with higher engagement, and consistent growth. @pranav_thampi_801e8cabe86

Some comments may only be visible to logged-in visitors. Sign in to view all comments.