AI-powered social listening platform built to cut through the noise
Impact of our work
プロジェクトについて
Project type
AI Social Listening Platform
Tech stack
Next.js tRPC Supabase OpenAI
built in
3 months
8 platforms
real-time social listening
8 AI categories
intelligent post classification
MVP launched
from idea to production in 3 months
About the project
プロジェクト
Thredly is an AI-powered social listening platform that helps brands instantly surface the conversations that actually matter. Instead of endless scrolling or blunt keyword alerts, Thredly uses AI to reduce noise and categorize posts into 8 high-signal buckets: Questions, Recommendations, Technical Support, Brand Mentions, High-Intent Sales, Competitor Mentions, Reviews and Feedback, and Complaints and Issues.
Thredly supports listening across X, YouTube, TikTok, Instagram, Reddit, LinkedIn, Facebook, and Threads, giving teams a single place to understand what people are really saying across platforms.
The challenge
挑戦
ID.5.001
Turn an ambitious AI product idea into a focused, usable MVP quickly, with intelligent matching across 8+ social platforms and real-time categorization that goes beyond simple keyword matching.
The solution
解決策
ID.1.001
AI categorization pipeline
Built an intelligent monitoring system that generates search queries based on brand context, scores relevancy using AI, and classifies posts into 8 high-intent categories.
ID.1.002
Multi-platform integration
Connected 8+ social media APIs with OAuth and platform-specific data normalization.
ID.1.003
tRPC + Supabase
Type-safe API layer with Supabase for PostgreSQL, real-time data, and Row-Level Security.
ID.1.004
Background job processing
Scalable monitoring pipeline with batch processing, concurrency controls, and GCP integration for distributed execution.
ID.1.005
AI response generation
Context-aware, on-brand response drafting using organization-specific brand knowledge and tone guidelines.
ID.1.000
BYLD shaped, designed, and shipped Thredly end to end. We defined the product strategy and feature scope, then built the full-stack application on a modern monorepo architecture with Next.js, tRPC for type-safe APIs, and Supabase for data and auth.
The AI pipeline uses Vercel AI SDK with OpenAI and Grok models to generate platform-specific search queries, score post relevancy, and classify conversations into 8 intent categories. A background monitoring system continuously scans connected platforms, processes posts in batches with concurrency controls, and delivers digest emails with the top signals.
We integrated Stripe for per-platform subscription billing, Clerk-based OAuth for social platform connections, and built a multi-tenant organization system with role-based access.