Casto: AI Content Repurposing Agent

Overview

Casto is an AI-powered content repurposing agent built with v0 that transforms blog posts into platform-optimized content for X, LinkedIn, Newsletter, and Reddit. Features a real agentic workflow with analysis, generation, evaluation, and refinement steps.

Features

• Real Agentic Workflow: 4-step pipeline (analyze → write → evaluate → refine)
• Multi-platform Support: X, LinkedIn, Newsletter, and Reddit
• Real-time Visualization: Watch AI thinking through chain-of-thought display
• Content Dashboard: Save, manage, and track all generated content
• Engagement Tracking: X API integration for automatic metrics
• Notion Integration: Save drafts directly to Notion workspace

What Users Can Do

• Transform Content: Turn blog posts into platform-native posts
• Watch AI Work: See real-time progress and decision-making
• Manage Content: Access dashboard with all generated content
• Track Engagement: Monitor performance across platforms
• Save to Notion: Export content with one click
• Customize Output: Platform-specific optimization

Why I Built This

As a content creator, I faced the challenge of adapting long-form content for different platforms. Each platform has unique requirements - character limits, tone, formatting, and audience expectations. Manually creating platform-specific versions was time-consuming and inconsistent.

Casto solves this by implementing a genuine agentic workflow that:

• Understands content intent and audience first
• Generates platform-specific versions with best practices
• Evaluates its own output for quality
• Refines content based on evaluation feedback

Built entirely with v0 by Vercel to showcase what's possible with AI-native development.

Tech Stack

• Framework: Next.js 15 (App Router)
• AI: Vercel AI SDK for agentic pipeline
• Database: Supabase (PostgreSQL)
• Authentication: Supabase Auth
• Styling: Tailwind CSS + shadcn/ui
• APIs: X API v2, Notion API
• Deployment: Vercel

Key Components

Agentic Pipeline

Step 1: Analyze

• Understands content intent, tone, and audience
• Extracts core message and key points
• Recommends approach for each platform

Step 2: Write

• Generates platform-specific content
• Follows best practices and character limits
• Adapts tone for target audience

Step 3: Evaluate

• Automatically assesses quality
• Identifies issues and strengths
• Determines if refinement is needed

Step 4: Refine

• Conditionally polishes content
• Fixes identified issues
• Preserves strengths

Usage

1. Paste your blog content into the editor
2. Select target platforms (X, LinkedIn, Newsletter, Reddit)
3. Click "Run Agent" and watch the workflow
4. Review AI-generated platform-specific content
5. Save to dashboard or export to Notion
6. Track engagement metrics over time

Architecture

• Frontend: Real-time streaming UI with Server-Sent Events
• Backend: API routes for content generation and management
• AI Processing: Structured generation with evaluation loops
• Database: Row Level Security for user content
• Integrations: X API for metrics, Notion for exports

Key Features

• Chain-of-Thought Visualization: See AI reasoning in real-time
• Conditional Refinement: Only refines when evaluation finds issues
• Session Memory: Learns from feedback within sessions
• Multi-model Support: 18+ AI models via Vercel AI Gateway
• Quality Scoring: Automatic evaluation of generated content
• Post Tracking: Link published posts and track performance

Impact

Built as a v0 Ambassador to demonstrate production-ready AI applications with proper agentic patterns. The template showcases:

• Real agentic workflows (not just prompt wrappers)
• Structured AI outputs with quality evaluation
• Real-time progress streaming
• Full-stack integration with auth and database
• External API integrations (X, Notion)

Perfect for content creators, marketers, and developers building AI applications.