Three weeks ago I launched Vestron because Instagram has ignored a massive UX problem since 2016 and I got annoyed enough to build something about it.
The Problem
Instagram lets you save posts and reels but gives you zero way to search them. Collections become completely unusable once you have more than 50 saves. I personally had 600+ saved reels and spent 12 minutes trying to find a cake recipe for my mom's birthday. Never found it. That was the breaking point.
What I Built
Vestron is a mobile app that organizes Instagram saves automatically using AI. You share posts to an Instagram bot, AI categorizes everything, you search by describing what you remember, and it finds it in seconds.
Tech Stack
Frontend: React Native (Android live, iOS planned)
Backend: Node
AI: Gemini API for semantic search and categorization
Integration: Instagram Graph API (Meta verified)
Auth: Bot-based (no Instagram login required)
Why Bot-Based Auth?
Instead of traditional OAuth, users share posts to @vestron_bot like DMing a friend. This removes friction and addresses privacy concerns since we only see explicitly shared content. No access to DMs, followers, or profile data.
The Meta Verification Journey
Getting Meta verified was brutal. Full code review, privacy policy audit, security compliance documentation. Took about 3 weeks of back and forth. But it was necessary to use official APIs instead of scraping.
Technical Challenges
Semantic Search: Users rarely remember exact captions. "Italian pasta with cherry tomatoes" needs to find the reel even if the original caption was completely different. Claude API handles this surprisingly well.
React Native Performance: Some users have 100+ saves. Implemented virtualized lists and lazy loading to keep the app smooth.
Rate Limits: Meta's Graph API has strict rate limits. Had to implement smart caching and batch processing.
Current Numbers (3 weeks)
94 users, 1,124 saves organized, 12 saves per user average
Retention seems solid but still early to draw conclusions.
What Actually Worked for Distribution
Building the product was easier than getting users. Zero marketing budget. Here's what I've tried:
❌ LinkedIn posts (40 impressions, 0 downloads)
❌ Reddit (karma too low for most subs)
❌ Cold creator outreach (19% response, 0% conversion)
✅ Instagram reels showing my face (1,300 views, 3 downloads)
Turns out people don't want to hear about "AI-powered semantic search." They want to hear "I lost a recipe and got annoyed."
Why Mobile First?
Desktop tools exist (Instavault, Dewey) but they're Chrome extensions. Instagram is 80% mobile usage. Nobody's using Instagram on desktop to save content. Mobile first seems obvious in hindsight.
Open Questions
For those who've built on Meta's APIs: how's your experience with API stability and rate limits?
Anyone solved similar mobile app distribution challenges with zero budget?
The app is free on Android: vestron.in
Happy to answer technical questions about the Instagram Graph API integration or bot architecture.
Building in public. Learning as I go. 6 users away from 100 and that milestone is driving me insane.
Top comments (0)